Home

Jess in Action e-book

1. When you make a copy of the template you should replace with a description of your test and put the test code into test something The test code should dis play relevant results using printout so they can be compared to the reference file You need to create test ref as well One way to do this is to run the test without it of course the test will fail Carefully check over the test out file to make sure the output is correct If it is simply copy it to test ref If it is not but you understand what the correct output should be you might want to make a copy anyway and edit it to create a correct version in test ref The Java template The Jess template is for tests that consist entirely of scripts and the Java template is for tests written as Java code Otherwise the functionality is very similar The runtest sh for JavaTemplate looks like this The Java template 431 shelldefs mk echo Building make clean gt dev null 2 gt amp 1 make gt test errs 2 gt amp 1 echo Build done if test f test class then echo Build failed cat test errs exit else echo Build Succeeded rm test errs fi echo Running the test program S JAVA test gt test out 2 gt amp 1 if diff ignore space change test out test ref gt dev null then files are the same echo Test succeeded else files are different echo Test failed Try echo diff ignore space change test out test ref fi The only sub
2. do NOT use an old value when notifying of changes to FuzzyValues firePropertyChange will not pass it on since FuzzyValues with the same FuzzyVariable are considered to be EQUAL even if they have different FuzzySets m_pcs firePropertyChange fuzzyReading null Object m_fuzzyReading 272 oO 27 CHAPTER 16 Writing the rules try Thread sleep 1000 catch InterruptedException ie return This part of the code Creates a FuzzyVariable to represent the temperature and loads it with appropriate fuzzy terms like cool warm and so on The fuzzy reading is the value you match in your new set of rules so it needs to be available as a JavaBeans property This block of code sends out a property change notification for the fuzzy temper ature reading whenever a change is detected in the crisp temperature reading Next you augment the vent bean in a similar fashion to define the terms for con trolling the vent openings fullyClosed slightlyOpen halfOpen mostlyOpen and fullyOpen and also to notify Jess when the FuzzyValues change Again the modified code see listing 16 2 is highlighted in bold figure 16 3 shows the fuzzy sets for the new terms package fuzzy import nrc fuzzy public class Vent extends BeanSupport private Hardware m_hardware Create FuzzyVariable to private int m_floor represent vent position private double m_oldState private static FuzzyVariab
3. The Purchase servlet 349 for int i 0 i lt items length i Fact item new Fact line item engine item setSlotValue order number orderNumberValue item setSlotValue customer id customerIdValue item setSlotValue part number new Value items i RU ATOM engine assertFact item Iterator result engine runQuery items for order new ValueVector add orderNumberValue request setAttribute queryResult result catch JessException je throw new ServletException je dispatch request response purchase jsp Listing 19 3 purchase JSP lt HTML gt lt page import jess gt lt jsp useBean id queryResult class java util Iterator scope request gt lt HEAD gt lt TITLE gt Thank you for your order lt TITLE gt lt HEAD gt lt BODY gt lt H1 gt Thanks for shopping at TekMart com lt H1 gt These are the items you are purchasing If this were a real web site I d be asking for your credit card number now lt P gt lt TABLE border 1 gt lt TR gt lt TH gt Name lt TH gt lt TH gt Catalog lt TH gt lt TH gt Price lt TH gt lt TR gt lt double total 0 while queryResult hasNext Token token Token queryResult next Fact fact token fact 2 double price 350 CHAPTER 19 Deploying web based applications fact getSlotValue price floatValue null total price gt lt TR gt lt TD gt lt
4. Ifa customer is buying a videotape or DVD recommend another one If a customer has bought a videotape or DVD in the past recommend another one In every case you should keep track of previous purchases so you don t recom mend a videotape or DVD the customer has already bought from you Designing data structures After talking with the experts you find that the important entities in the system include the following Products Things that are for sale Customers Current and past users of the system Orders Current and past customer purchases Line items A collection of items that form an order Recommendations What the system produces as output 294 CHAPTER 17 Jess on the Web You can define a deftemplate for each item in this list The product template should include slots for the name catalog number and price of the product of course It also should include a slot to specify the category the product belongs to and the categories of other products required by this product deftemplate product slot name slot category slot part number slot price multislot requires For example a product named TekMart 19 TV might have the category tv and the requires slot could contain batteries for the remote control A customer is just a person with a name address and customer ID deftemplate customer multislot name multislot address slot customer id An order is a collec
5. answer ident sound text no answer ident plugged in text no gt recommend action plug in the computer halt defrule MAIN power supply broken declare auto focus TRUE answer ident sound text no answer ident plugged in text yes gt recommend action repair or replace power supply halt _ Figure12 2 and all the flowcharts you ll use to develop this application are vastly simplified The action Check power supply should ideally expand to a whole flow chart of its own explaining the steps involved in diagnosing power supply prob lems Unfortunately to present that flowchart and all the rules to support it and the dozens of others you d need to collect I d fill another whole book Because this is a programming book not a PC repair manual I ll keep things simple These rules assume something else is taking care of asking the user ques tions the rules just depend on the answers Whereas in the Tax Forms Advisor application from part 3 you had to explicitly write rules to ask the questions in this application the questions are asked automatically using backward chaining START Does computer oot YES Check plugged in power supply Goto sound flowchart Figure 12 2 A simple top level flowchart for diagnosing PC hardware problems This flowchart can diagnose the problem The computer isn t getting any
6. return which pump public Value call ValueVector vv Context c throws JessException int floor vv get 1 intValue c int heatPump floor 1 3 1 return new Value heatPump RU INTEGER This basic Userfunction should be familiar to you by now The HVAC functions 251 15 5 4 Operating the hardware You need several Userfunctions to read and change the operating mode of each heat pump and to check open and close the vents It would be nice if you could use symbolic names for the heat pump and vent states something like this Jess gt get hp state 2 cooling Jess gt set vent state 4 open This is easy enough to implement Because there are so few alternatives the call methods can contain multiway branches or switch statements to handle each of the possible options The function to read the heat pump state looks like this package control import jess class ReadHeatPump implements Userfunction public String getName return get hp state public Value call ValueVector vv Context c throws JessException int heatPump vv get 1 intValue c Simulator sim InitSimulator getSimulator c int state sim getHeatPumpState heatPump try return new Value State heatpump state RU ATOM catch IllegalArgumentException iae throw new JessException get hp state Unexpected state state This Userfunction returns symbols values of type RU ATOM to represent the hea
7. 20 4 1 An example rule Using this DTD you can easily write many Jess rules in XML A rule like this defrule AnimalRule2 declare salience 10 animal lt animal has hair TRUE gt modify animal type mammal can be represented as shown by the XML document in listing 20 1 Representing Jess rules in XML lt xml version 1 0 gt lt DOCTYPE rulebase SYSTEM jess dtd gt lt rulebase gt lt rule name AnimalRule2 priority 10 gt lt lhs gt lt pattern name animal binding animal gt lt slot name has hair gt lt constant gt TRUE lt constant gt lt slot gt lt pattern gt lt lhs gt lt rhs gt lt function call gt lt head gt modi fy lt head gt lt variable gt animal lt variable gt lt constant gt type mammal lt constant gt lt function call gt lt rhs gt lt rule gt lt rulebase gt 369 20 4 2 Transforming the XML rules into Jess rules The XML rule format can be transformed into Jess rules using an XSLT script of about 100 lines XSLT programs are declarative rather than procedural just like rules in Jess In fact an XSLT program is precisely a list of rules for transforming specific parts of an XML document into some desired result format This isn t a book about XSLT programming so I won t go over the code and try to explain it but because XSLT is rule based it seems appropriate to include this script here see listing 20 2 along with a b
8. The formula takes the difference of each of the X Y and Z coordinate pairs adds them and takes the square root It is long and messy and you surely don t want to type it more than once Many different rules need to do the same computation so you can define a function named distance and call it from each rule as needed 56 CHAPTER 4 Getting started with the Jess language You can write your own functions in the Jess language using the deffunction construct Once you define a deffunction you can use it like any other Jess func tion deffunction looks like this deffunction lt name gt lt parameter gt lt comment gt lt expression gt The name of the deffunction must be a symbol Each parameter must be a vari able name complete with the question mark A function can have as many param eters as you need to use The optional comment is a double quoted string that can describe the purpose of the function This special comment is included when a deffunction is dis played It s a good idea to provide a comment that succinctly describes the pur pose of the function The comment like any Jess string can span multiple lines The body of a deffunction is composed of any number of expressions The return value of the deffunction is the value of the last expression evaluated unless you use the return function to provide an explicit return value Here s what the distance calculation looks like as a deffunction Jess gt
9. The same variable can appear in more than one pattern in the same rule or in several places within one pattern or both Every time the variable is used it must match the same value In listing 7 2 although two facts could match each pattern individually only one pair can activate the rule the fact a 2 and b 2 Jess gt defrule repeated variables a x b x gt printout t x is x crlf Constraining slot data 103 TRUE Jess gt watch activations TRUE Jess gt deffacts repeated variable facts a 1 TRUE Jess gt reset gt Activation MAIN repeated variables f 2 f 3 TRUE Jess gt run 2238 2 1 E Note that in Jess 6 1 you can t use a variable to match the head of a fact or the name of a slot these things must always be specified as literal values This capabil ity is planned for a future release however Multifields Regular variables match exactly one value Multifields can match any number of values zero one or more You write a multifield by preceding a variable name with the characters for example mf is a multifield You can only use multi fields in multislots They can be used alone in which case the multifield matches any number of values in that multislot or in combination with regular variables or literal values If you use multifields together with single values the multifields expand to match everything not matched by the other values For example the
10. declare salience 100 command exit when idle gt printout t exiting crlf TRUE This rule won t fire until no other rules of higher salience are on the agenda Declaring a low salience value for a rule makes it fire after all other rules of higher salience A high value makes a rule fire before all rules of lower salience The default salience value is 0 so if this is the only rule with an explicit salience value it will not fire until the agenda is empty You can specify salience values using literal integers global variables or func tion calls How the salience values are evaluated depends on the current value of the salience evaluation method These values are as follows m when defined Default A fixed salience value is computed when the rule is defined m when activated tThe salience of a rule is reevaluated each time the rule is activated m every cycle The salience value of every rule on the agenda is recom puted after every rule firing Evaluating every cycle is very computation ally expensive and isn t used much You can query or set the salience evaluation method with the set salience evaluation and get salience evaluation functions Note that extensive use of salience is generally discouraged for two reasons First use of salience has a negative impact on performance at least with the built in conflict resolution strategies Second it is considered bad style in rule based programming to
11. eat Cereal Spoon END The cereal program with its predictable linear control flow is typical of many small computer programs Calculating the value of an equation computing a customer s bill at a shopping mall even rendering a single frame of a complex video game are all variations on the same sort of linear deterministic process in which each step follows inevitably from the last Such programs are often called procedural programs they solve a problem in a straightforward predictable way Traditional software development techniques can be used to write programs such 2 1 1 The cooking driving robot 15 as the cereal program to good effect You wouldn t write a rule based program to solve the cereal problem In general problems that have a well known algorith mic solution should be solved using a traditional procedural approach Emboldened by your success you now think about teaching the same robot how to drive a car in the real world You have to tell the robot to search for its keys and driver s license but only if it isn t already carrying them You tell the robot how to start the car put the car in reverse and back out of the garage but only if the garage door is already open The instructions probably need to cover the different behaviors of cold motors and warm motors how to use both man ual and automatic transmissions and different types of emergency brakes And the instructions will become far more compl
12. engine executeCommand assert numbers 2 2 String rule defrule add numbers numbers n1 n2 oyn 4 store SUM n1 n2 Use executeCommand to define a rule engine executeCommand rule The rule fires and stores the result engine executeCommand run Value sumValue engine fetch SUM int sum sumValue intValue engine getGlobalContext The Java variable sum is initialized to 4 This is a long winded way of adding two numbers of course but the same technique works with more complex rules Beyond executeCommand As you ve seen you can use the executeCommand method to do just about any thing in Jess from Java This approach has some problems though It s verbose Beside typing the whole Jess command you have to type exe cuteCommand and possibly some store and fetch calls It s potentially inefficient because Jess has to parse the argument If the script passed to executeCommand is something that will take a while to run this issue doesn t matter but for a single short command it could be a lot of overhead It s error prone If the Jess script contains a syntax error the Java compiler won t catch it you won t find the error until you run your program The executeCommand method is very convenient then but it has some important disadvantages Whenever possible you should consider using Jess s direct Java APIs instead Table 18 1 lists some Jess fun
13. explanation Income below threshold no dependents Similarly Form 1040A can be filed if the user s income is below the same thresh old but she has either dependents or excess interest income or both You can write this as two rules defrule form 1040A excess interest user income i amp lt i 50000 answer ident interest text yes gt assert recommendation form 1040A explanation Excess interest income defrule form 1040A user income i amp lt i 50000 dependents d amp gt d 0 gt assert recommendation form 1040A explanation Income below threshold with dependents The U S Federal Tax Code is famously complicated It s common practice for U S residents to try filling out several different combinations of tax forms because each combination can result in a slightly dif ferent calculated tax liability The taxpayer then files the combination of forms that leads to the smallest tax bill Recommending forms 179 You recommend the long Form 1040 if the user needs to itemize deductions or if the user s income is above the threshold The rule to handle this latter condition is simply defrule form 1040 income above threshold user income i amp gt i 50000 gt assert recommendation form 1040 explanation Income above threshold The rest of the rules in this module handle the various special circumstances The user can take the
14. firstLoopDone reading m_oldReading 0 0 boolean readingChanged reading m_oldReading m_pcs firePropertyChange reading new Double m_oldReading new Double reading m_oldReading reading if readingChanged try m_fuzzyReading new FuzzyValue m_thermometerFV new SingletonFuzzySet reading catch XValuesOutOfOrderException xvorder System out println Error xvorder return catch XValueOutsideUODException xvuod Update F variable System out println Error xvuod return representing rate of m_pcs firePropertyChange fuzzyReading temperature null m_fuzzyReading change try m_fuzzyReadingRateOfChange new FuzzyValue m_thermometerRateOfChangeFV new SingletonFuzzySet rateOfChange catch XValuesOutOfOrderException xvorder System out println Error xvorder return catch XValueOutsideUODException xvuod 27 Controlling with fuzzy rules 279 System out println Error xvuod return m_pcs firePropertyChange fuzzyReadingRateOfChange null1 m_fuzzyReadingRateOfChange firstLoopDone true try Thread sleep 1000 catch InterruptedException ie return The code sets up a FuzzyVariable to represent the rate of change of temperature readings with the three values decreasing increasing and zero You make the temperature change rate variable available as a JavaBeans property This new method lets
15. Finally note that literal constraints have to match exactly no conversions are done Thus the floating point literal 1 0 doesn t match the integer 1 and the sym bol xyz doesn t match the string xyz This is a common source of problems when using shadow facts see section 6 5 Variables as constraints If all the patterns of a rule had to be given literally Jess would not be very power ful However patterns can also include variables and various kinds of predicates comparisons and boolean functions and can be modified by conditional elements We ll consider variables and predicates here and conditional elements in the fol lowing sections You can specify a variable instead of a literal value for any part of the slot data in a pattern A variable matches any value in that position within the facts that match that pattern For example the rule Jess gt defrule simple variables a x Py gt printout t Saw a 9x y erlt is activated each time an ordered fact with head a having two fields is asserted a b c a1 2 aaa and so forth The variables thus matched on the LHS of a rule are available on the RHS of the same rule you can think of them as the arguments to the rule s RHS when it fires For example if the previous rule matched the fact abc then when the rule fired x would have the value b and y would have the value c You can mix literal values and variables in the same pattern of course
16. Generally look as though you re happy to be talking to the person or they won t talk to you again Dress for the occasion Gone are the days when all white collar workers wore white collars and ties But if you re interviewing someone older than you she might remember those days quite clearly If you re going to interview a client at a bank don t show up in sandals and a t shirt Dressing appropri ately will help your interviewee relate to you Introduction to knowledge engineering 155 Be reassuring Often the interviewee is not the customer A manager may be asking you to capture knowledge from an employee and that employee may be afraid of being replaced by the proposed new system Reassure the employee that he s smarter than any computer and explain that although the system may take over routine tasks it will free the employee s time to work on more important things You don t want anyone to perceive you or the system you re building as an enemy Customers The customers are the people who are paying you to build the system Sometimes they know a lot about the problem and other times they just want the problem solved If the customer is also a domain expert then your job is easy because the customer can direct you to all the information you need If the customer doesn t know much about the problem domain then the hardest part of your job may be identifying someone who is For the forms advisor appl
17. Now let s write a rule that calculates the factorial of a number when it is needed The rule should directly match the need factorial trigger facts Jess gt defrule do factorial need factorial x gt compute the factorial of x in r bind r 1 bind n x while gt n 1 bind r r n bind n n 1 assert factorial x r TRUE The rule compiler rewrites rules like this too It adds a negated match for the factorial x pattern to the rule s LHS so the rule won t fire if both the goal fact and the corresponding goal seeking fact are both present The end result is that you can write rules that match on factorial facts and if they are close to firing except they need a factorial fact to do so any need factorial rules may be activated If these rules fire then the needed facts appear and the factorial matching rules fire This then is backward chaining Note that any needed factorial facts are created only once so the expensive computa tion need not be repeated Often avoiding redundant computation is one of the main benefits of backward chaining Jess chains backward through any number of reactive patterns In the example in listing 7 4 imagine you have a database that allows you to look up the price of an item given its item number or the item number given its name To find the price given the name you need to do two separate queries When the price check fact is first asserted none o
18. Returns the contents of this value as a list jess ValueVector 425 public long longValue Context c Returns the contents of this value as a long public double numericValue Context c Returns the contents of this value as a number public String stringValue Context c Returns the contents of this value as a string public String variableValue Context c Returns the contents of this value as a string a variable name B 12 jess ValueVector A ValueVector is a list of Value objects This class is used both as is and also as base class for Fact and other important Jess classes ValueVector is very similar to the java util Vector and java util ArrayList classes It has separate notions of a capacity and a size A ValueVector contains an array of Value objects The capacity is the length of the array the size is the number of Value objects actually in the array The capacity is automatically expanded if you use add to add Value objects public ValueVector Constructs a ValueVector with the default capacity 10 public ValueVector int size Constructs a ValueVector with the given capacity public ValueVector add Value val Adds a new element to the end of this ValueVector public Value get int i Fetches the entry at position i in the list public ValueVector remove int i Removes the item at the given index from the list moves all the highernumbered elements down public ValueVector set Value val int i Sets th
19. deffunction distance X1 Y1 Z1 X2 Y2 Z2 Compute the distance between two points in 3D space bind x X1 X2 bind y Y1 Y2 bind z Z1 22 bind distance sqrt x x y Py z z return distance TRUE You can now call distance just as if it were built in to Jess Jess gt distance 10 0 0 2 0 0 8 Besides being shorter and easier to type the function call is much easier to under stand the reader immediately knows what s being computed instead of needing to puzzle out those subtractions multiplications additions and square root functions The distance function computes a single value More complex deffunctions can use the control flow functions like foreach if and while that you learned about in section 4 3 Your video game rules will need to make decisions based on which of several rival ninjas is closest to the player The following example is a deffunction that returns the smaller of its two numeric arguments Jess gt deffunction min a b if lt a b then return a else return b TRUE 4 4 1 Defining functions with deffunction 57 Note that this could also be written as Jess gt deffunction min a b if lt a b then 2a else 2b TRUE because the expressions in the body of a deffunction do not have to be function calls and a deffunction returns the value of the last expression evaluated One limitation of min is that it takes only two
20. is the same as and not x not y not and x y is the same as or not x not y Jess does this rearrangement so that it can form subrules which are the topic of the next section Subrule generation and the or conditional element A rule containing an or conditional element with n branches is precisely equiva lent to n rules each of which has one of the branches on its LHS In fact this is how Jess implements the or conditional element Jess internally divides the rule generating one rule for each branch Each of these generated rules is a subrule For a rule named rule namne the first subrule is also named rule name the sec ond is rule name amp 1 the third is rule name amp 2 and so on Each of these subrules is added to the Rete network individually If you execute the rules command which lists all the defined rules you will see each of the subrules listed separately If you use the ppdefrule function to see a pretty print representation of a sub rule you will see only the representation of that particular subrule Note that because amp is a token delimiter in the Jess grammar you can only refer to a subrule 7 3 3 Qualifying patterns with conditional elements 111 with an ampersand in the name by placing the whole name in quotes for exam ple ppdefrule rule name amp 6 Jess knows that the subrules created from a given rule are related If the origi nal rule is removed either using undefrule or im
21. lt BODY gt lt HTML gt 19 12 The Purchase servlet The Purchase servlet is a combination of the other two servlets it makes sure the customer is logged in asserts any items request parameters as line item facts runs the items for order query and then passes the Iterator along to the pur chase JSP In a full fledged e commerce system the purchase JSP would collect shipping and credit card information here it just displays the items and sums their prices The Purchase servlet is shown in listing 19 2 and the JSP is in listing 19 3 Listing 19 2 Purchase servlet public class Purchase extends BaseServlet public void doGet HttpServletRequest request HttpServletResponse response throws IOException ServletException checkInitialized ServletContext servletContext getServletContext String orderNumberString String request getSession getAttribute orderNumber String customerIdString String request getSession getAttribute customerId if orderNumberString null customerIdString null dispatch request response index html1 return try Rete engine Rete servletContext getAttribute engine int orderNumber Integer parseInt orderNumberString Value orderNumberValue new Value orderNumber RU INTEGER Value customerIdValue new Value customerIdString RU ATOM String items String request getParameterValues items if items null
22. s event thread and the main thread where rules fire In this and the previous part you developed complete isolated applications from the ground up using Jess In the real world however Jess usually needs to Summary 217 cooperate with Java code outside of itself In the next part of this book you will learn how to put Java objects into Jess s working memory so that Jess can reason about the outside world You ll also learn how to extend the Jess language by writ ing functions in Java Part 5 Reasoning about reality the HVAC Controller I he applications you ve written so far have been written entirely as Jess scripts In much real software however the rule engine has to be integrated with existing Java code In part 5 you ll write some Java as well as Jess code and in so doing you ll begin to learn about Jess s Java APIs The HVAC Controller you ll write is a soft real time control system for the heat ing and cooling systems in an office building We ll look at how such a system can be interfaced to real hardware You ll write rules that react not just to ordinary facts in Jess s working memory but to shadow facts that are tied to JavaBeans the JavaBeans will be wrappers around the HVAC interface software At the end of chapter 16 is a special essay by Bob Orchard the author of the FuzzyJess extension that adds fuzzy logic capabilities to Jess Bob looks at how the HVAC Controller can be extended and improved fu
23. t very interesting Combined with or and not conditional ele ments though you can use the and CE to construct complex logical conditions The or conditional element Any number of patterns can be enclosed in a list with or as the head The or CE matches if one or more of the patterns inside it matches If more than one of the patterns inside the or matches the entire or is matched more than once Jess gt clear TRUE Jess gt deftemplate used car slot price slot mileage TRUE Jess gt deftemplate new car slot price slot warrantyPeriod TRUE Jess gt defrule might buy car candidate lt or used car mileage m amp lt m 50000 new car price p amp lt p 20000 gt assert candidate candidate Jess gt assert new car price 18000 lt Fact 0 gt Jess gt assert used car mileage 30000 lt Fact 1 gt Jess gt run 2 The rule fires twice once for the new car and once for the used car In this rule only one of the two branches of the or conditional element will match at a time but the rule can be activated separately as many times as there are facts to match Each of the vehicles listed matches only one or the other of the branches For some activations the first branch of the or will match and for others it will be the second branch Note that the variable candidate is bound to whatever fact matches the or in each particular activation If might buy car s RHS tried to modif
24. 127 0 0 1 8080 HelloJess servlet HelloJess you should see much the same web page as before except this time it is generated by Jess Application architecture take one 331 19 3 2 Cleaning up the URL The URL you ve been typing to see the servlets is quite verbose By modifying the web xml file slightly you can tell Tomcat to shorten the URL considerably Specifically you can tell Tomcat that a request for any URL starting with http 127 0 0 1 8080 HelloJess should be routed to the HelloJess servlet includ ing for instance http 127 0 0 1 8080 HelloJess index html Before the closing lt webapp gt tag add the following block lt servlet mapping gt lt servlet name gt HelloJess lt servlet name gt lt url pattern gt lt url pattern gt lt servlet mapping gt After reinstalling the application and restarting Tomcat you can see the Jess servlet by visiting http 127 0 0 1 8080 HelloJess Note that now it s pretty much impossi ble to tell that the page is being generated by a servlet both the page itself and the URL look perfectly normal as if the page were a static HTML document 19 4 Application architecture take one Now that the preliminaries are taken care of and you know how to write simple servlets that embed Jess it s time to design and implement the Recommendations Agent As a first stab at a design for the application let s assume it will include four screens A login screen where
25. Calls fireProperty Change to brightness b indicate brightness property adjustTriac b is changing pes firePropertyChange brightness new Integer old new Integer b private void adjustTriac int brightness Code not shown private PropertyChangeSupport pcs new PropertyChangeSupport this public void addPropertyChangeListener PropertyChangeListener p Boilerplate code pcs addPropertyChangeListener p you can reuse in JavaBeans public void removePropertyChangeListener PropertyChangeListener p pces removePropertyChangeListener p Now if you use definstance without the static qualifier to register a Dimmer SwitchDynamic instance with Jess the shadow fact tracks the brightness property whenever it changes as listing 6 4 demonstrates This time calling setBrightness to change the JavaBean changes the shadow fact too Jess receives a Property ChangeEvent and modifies the shadow fact accordingly Listing 6 4 A demonstration of dynamic definstances Jess gt clear lt Remove old definitions TRUE Create new Jess gt defclass dimmer DimmerSwitchDynamic y dimmer template dimmer Jess gt bind ds new DimmerSwitchDynamic Create instance of lt External Address DimmerSwitchDynamic gt new JavaBean class Jess gt definstance dimmer ds If static isn t specified lt Fact 0 gt dynamic is default Jess gt facts f 0 MAIN dimmer brightness 0 class lt External Address jav
26. Executing the rule engine In the runtime phase you create a RuleSession A RuleSession is basically an instance of a rule engine with a specific set of rules loaded into it You get RuleSessions from the javax rules RuleRuntime object and you get the Rule Runtime in turn from the RuleServiceProvider 386 21 3 2 CHAPTER 21 Jess in the J2EE environment There are two kinds of RuleSessions statefuland stateless They differ in the capabilities they offer The working memory of a StatefulRuleSession per sists between method calls You can add objects to working memory in a series of invocations then run the engine and then add more objects and run again In contrast a StatelessRuleSession is a one shot affair To call its executeRules methods you must supply the entire initial contents of working memory and you will receive a list of the final contents as a return value Here you create a StatefulRuleSession add two objects an Integer anda String to working memory execute the rules and then retrieve the entire con tents of working memory as a java util List Finally you dispose of the RuleSession by calling release RuleRuntime runtime rsp getRuleRuntime StatefulRuleSession session StatefulRuleSession runtime createRuleSession rules properties RuleRunt ime STATEFUL_SESSION_TYPE session addObject new Integer 1 session addObject A string session executeRules List results session getObjec
27. In chapter 15 you ll even learn how to extend the Jess language by adding functions written in Java All code in Jess control structures 4 2 3 Adding some structure 47 assignments declarations takes the form of a function call You have already seen most of the Jess language s built in syntax virtually everything else is accom plished by functions that use these basic elements Let s look at some simple func tions that are used in most Jess programs In section 4 3 you ll learn about more complex functions that implement control structures One of the most commonly used functions is printout The printout function is used to send text to Jess s standard output or to a file The first argument tells printout where to send its output The subsequent arguments are printed one after another with no spaces between A complete explanation of the first argu ment will have to wait but for now all you need to know is that if you use the sym bol t printout sends its output to the console The special symbol cr1f is printed as a newline Here s an example of printing several arguments Jess gt printout t The answer is 42 crlf The answer is 42 Another useful function is batch which evaluates a file of Jess code To run the Jess source file examples hello clp which comes with the standard Jess distribu tion you can enter the following command Jess gt batch examples hello clp Hello World Another simple funct
28. It would not be reasonable to expect bload to be able to restore the control program to a consistent state from offline storage Grouping functions with Userpackage If you ve written a group of related Userfunctions either in a single class or in a group of classes it would be convenient to provide a way for them to be loaded together into Jess with a single function call This is especially true if your Userfunction objects need constructor arguments like the AnyString 248 15 5 15 5 1 CHAPTER 15 Extending the Jess language class earlier The jess Userpackage interface takes care of this for you User package has one method add you implement add to initialize and install any number of Userfunction objects Here s a simple Userpackage to install a few AnyString functions import jess public class AnyStringPackage implements Userpackage public void add Rete engine engine addUserfunction new AnyString fred engine addUserfunction new AnyString george engine addUserfunction new AnyString ron engine addUserfunction new AnyString percy You can install a Userpackage into Jess using the load package function or the addUserpackage method in the Rete class In the next section you will finally write the HVAC functions and collect them in a Userpackage The HVAC functions You ve learned everything you need to know to develop the Userfunctions for the HVAC system You need to be ab
29. The defclass function first discussed in section 6 5 tells Jess that you re going to define shadow facts for objects of a certain class here control Thermometer The definstance function installs an object of that class into Jess Finally the rule here matches only Thermometers that are reading over 72 degrees The tem peratures in the simulator start at 70 and drift up quickly so this rule fires after a short interval Notice how once a JavaBean is installed into Jess it looks just like an ordinary fact and the Thermometer pattern that matches the Bean here is just an ordinary pattern You use run until halt instead of run because you want Jess to wait for the rule to be activated the agenda will be empty when you call run until halt but when the temperature becomes high enough the rule will activate and fire Writing the JavaBeans 233 14 3 2 Writing the other Beans The Hardware interface like the C library it wraps represents the state of a vent as a Boolean value and the state of a heat pump as an integer Because symbols like open and closed would be easier to work with than the corresponding Bool ean values true or false it would be a good idea to write the Vent and Heat Pump JavaBeans to use meaningful strings as property values You need to convert both ways between these symbols and the underlying integer and Boolean values and you ll need to do so in the code you write in the next chapter too so let s isolate
30. The object is wrapped in a Value object before being stored public Value store java lang String name Value val Stores a Value under the given name The import table When you name a Java class in Jess the import table is consulted as necessary to figure out the fully qualified name These methods add to the import table and use it to resolve classes public java lang Class findClass String clazz Given the name of a class uses the import table to come up with the fully quali fied name If you pass String as the argument the Class for java lang String is returned public void importClass String clazz Adds a fully qualified class name to the import table public void importPackage String pack Imports every class in a package by adding the package name to the import table Finding things A whole set of methods is available for retrieving constructs like instances of Defrule and Deffacts from the engine They all accept the name of the con struct as an argument If the name doesn t include a module name the name of the current module is used This set of methods includes findDeffacts 422 B 8 5 APPENDIX B Abridged Java API for Jess findDefrule findDeftemplate and findUserfunction findDefrule can find Defquery objects as well as Defrules There are several methods for finding specific facts in the working memory public Fact findFactByFact Fact f Finds fast Can be used to find out quickly if a given fa
31. Unordered facts are work ing memory elements that behave like rows in a database table although the col umns are traditionally called slots Here are some examples of unordered facts person name Bob Smith age 34 gender Male automobile make Ford model Explorer year 1999 box location kitchen contents spatula The person fact has slots name age and gender the automobile fact has slots make model and year The head of an unordered fact person automobile box is like the table name and the slot names are like the column names Before you can work with unordered facts you have to specify their structure using the deftemplate construct Once you have created some unordered facts and put them in working memory functions such as modify and duplicate let you change or copy them The deftemplate construct Typically a relational database contains many tables one for each type of informa tion the database holds In a real relational database to create a new table you have to specify the names of the columns that will be found in the table Thereaf ter every row in the table has those same columns no more and no less Similarly in Jess you define many different kinds of unordered facts Before you can assert an unordered fact with a given head you have to use the deftemplate construct to define the slots that kind of fact This example defines an unordered fact type with three slots Jess gt de
32. You ll see how each system is architected and you ll see the detailed implementation of each one All the example systems use the Jess rule engine In this chapter we ll take a closer look at Jess itself Jess is an interpreter for the Jess rule language The syntax of the Jess rule language is similar to that of Lisp so it might look strange to you at first but Jess s language is simple easy to learn and well suited to both defining rules and procedural programming Although Jess is in some sense fairly small it s a very rich environment Even after you ve spent the next few chapters learning about Jess and its rule language there will be plenty more to learn on the fly as you develop the major applications Although previous experience with Lisp might help you begin to understand the Jess rule language it is not necessary this book explains all you need to know On the other hand I assume you have some familiarity with Java and I assume you have a Java environment installed and know how to use it to compile and run Java applications The Jess rule engine Jess http herzberg ca sandia gov jess is a rule engine and scripting lan guage developed at Sandia National Laboratories in Livermore California in the late 1990s It is written in Java so it is an ideal tool for adding rules technol ogy to Java based software systems The CLIPS expert system shell http www ghgcorp com clips CLIPS html1 an open source rule e
33. You should run all the tests you have written quite often ideally each time a change is made to the system When the final system is delivered the tests can be Developing rule based systems 25 part of the deliverable they will be a great help to anyone who needs to modify the system in the future How do you develop tests Some tests will be very small and check intermedi ate results whereas others will be fully worked problems In the former case you might develop the test by yourself The larger tests though should be based when possible on actual case studies of how problems were solved in the past It is important that the tests be automated so no human checking of results is required otherwise the tests will require too much effort to run and will not be used It helps to have an automated test framework you can often quickly develop one yourself using Perl shell scripts or similar scripting facilities This testing technique known as test driven development is one facet of eXtreme Pro gramming a methodology that is rapidly gaining acceptance in many computer programming fields An automated test framework that I use for testing Jess pro grams is described in appendix C 2 4 4 Interface building For most rule based systems to do any useful work they need to be connected in some way to their environment Sometimes this means database access other times it means directly reading values from sensors and sending comman
34. a Learn how Jess compiles rules a Be introduced to the Rete algorithm a Use the view and matches functions a Learn how to make your rules more efficient 133 134 8 1 CHAPTER 8 Under the hood how Jess works You can drive a car without understanding anything about what s under the hood But if you re driving through a desert miles from a phone and your engine starts to sputter and cough a little knowledge and a full toolbox could go a long way So it is with much of the software you use as a developer As long as everything is working well you can get by without much knowledge of its internal workings When something starts to go wrong however you may need extra knowledge and a few tools to fix it This chapter gives you both as far as Jess is concerned We ll look at the Rete algorithm the technique that Jess uses to do fast pattern matching You ll also learn about some tools built into Jess that let you kick the tires and look under the hood Some of the information in this chapter is generally true of any system based on the Rete algorithm and a little is specific to one version of Jess I ll try to make the distinction as we go along In general the version specific parts of Jess are in nonpublic Java classes so you d have to go out of your way to use them If you do though consider yourself warned that they are internal implementation details and any Java code you write that uses them may well break each ti
35. along with a rule to construct the user fact once the necessary information is available The rules request income and request num dependents have no patterns on their left hand sides LHS they are acti vated unconditionally defmodule interview defrule request income gt assert ask income These rules have no patterns and are activated defrule request num dependents by a call to reset gt assert ask dependents defrule request interest income If the total income is less than 50000 answer ident income text i amp lt i 50000 and there are no dependents answer ident dependents text d amp eq d 0 gt assert MAIN ask interest defrule assert user fact answer ident income text i answer ident dependents text d gt assert user income i dependents d 176 CHAPTER 11 Writing the application 11 2 2 Dealing with special circumstances The rest of the interview rules ask the user about special circumstances in her life during the past year In general if some of these special circumstances apply the user may want to fill out the long form and take the corresponding deduc tions even if her income qualifies her to use one of the short forms These special circumstances include moving and unreimbursed employee expenses Some of the questions about special circumstances are again asked conditionally based on other answers for example
36. and all of your program s rules and facts belong to it You can get a reference to this jess Rete object using the engine function in Jess The code engine run is therefore more or less equivalent to run Now let s return to our discussion of coordinating multiple threads in your appli cation The function you re interested in is called waitForActivations If this method is called when the current module s agenda is empty it doesn t return right away Rather it uses the wait method from the java lang Object class to wait for new activations to arrive Note that the only way waitForActivations can return is for code on some other thread to call a function that modifies work ing memory You can call waitForActivations from Jess like this engine waitForActivations Listing 13 3 shows the changed parts of a new version of the ask module that imple ments this idea Whereas the old version of ask contained just one rule that asked the question and returned the answer this new version contains two rules one that sets up the question ask question by id and one that validates the answer and either returns the answer or asks the question again collect user input deffunction ask user question type valid Set up the GUI to ask a question bind s question if eq type multi then bind s str cat s crlf Valid answers are foreach Pitem valid bind s str cat s Pitem bind s str
37. basically your Userfunc tion s handle to Jess itself You ll learn about Jess s ValueVector Value and Con text classes in the following sections A very simple implementation of call for kill might look like this public Value call ValueVector vv Context context System exit 1 return null The kill function kills the program and returns an error status to the operating system it s similar to Jess s built in exit function which calls System exit 0 This call method doesn t really return a value to Jess because the method never returns although the Java compiler insists you return something so I used nu1l Loading a Userfunction into Jess You ve written a Userfunction of your own containing the two methods described so far and you d like to try it out Let s say you ve written a class named org me Kill that implements the Userfunction interface using the two methods You can load it into Jess from the command prompt using the load function function 240 15 2 15 2 1 CHAPTER 15 Extending the Jess language Jess gt load function org me Kill TRUE Then you can call it like any other Jess function Jess gt kill The symbol is a system prompt kill exits to the operating system From Java you can create an instance of your Userfunction class and install it using the addUserfunction method in the jess Rete class kill is OK for a first attempt but it doesn t let you specif
38. call token fact 1 bind name fact slot value fact name printout t name crlf leather gloves mohair sweater FALSE Here you re looking for gifts between 20 and 100 and the query finds mohair sweater and leather gloves Let s break down this code to see what it s doing As previously stated run query returns the query results as a Java java util Iterator object The Iterator interface has a method next that you call to retrieve each individual result it also has a hasNext method that returns true as long as there are more 130 7 7 1 7 7 2 CHAPTER 7 Writing rules in Jess results to return That explains the while it hasNext call it next control structure it steps through each of the results returned by the query Each individual result is a jess Token object A Token is just a collection of jess Fact objects each Token holds one match for the query You call the fact method of jess Token to retrieve the individual jess Fact objects within the Token Each match begins with an extra fact a query trigger fact that initiates the matching process it is asserted by the run query command this fact is retracted automatically after the query is run Hence the argument to the call to fact is 1 not 0 Once you have the right fact you use the fact slot value function to extract the contents of the name slot Printing the name slot of each fact leads to the output shown earlier The defquer
39. defrule cooling and temp cool rate zero HeatPump state cooling number p Thermometer floor amp eq p which pump f fuzzyReading t amp fuzzy match t cool fuzzyReadingRateOfChange tr amp fuzzy match tr zero gt assert fuzzy vent floor f fuzzy change state new FuzzyValue ventFVar NS There will be one rule like this for each cell in table 16 4 Now you have 30 rules that determine how the vent should be changed You might need to think about these rules to see if they make sense to you but with a bit of analysis of the various situations the rules should be similar to what your intuition would lead you to As in the last example a single rule is defined with a lower priority so that it may take the combined output of all the rules and defuzzify the result giving the required change for the vent This rule in this second fuzzy controller is as follows defrule perform vent change declare salience 100 fv lt fuzzy vent floor f fuzzy change state fs gt bind vent change amount fs weightedAverageDefuzzify change vent state f vent change amount retract fv The only other major change is to the WriteVent class and the Simulator class to support the ability to change the vent state by the specified amount The function change vent state is a Userfunction defined in WriteVent java it in turn calls the simulator to calculate the new position for the vent a
40. downloading 33 embedding 222 formatting code 45 formatting constructs 419 functions 47 home page 32 installing 33 Java integration 62 library 307 license xxii parser 417 performance 38 39 rule language 32 42 running 34 user s manual 33 web site 33 Jess functions calling from Java 315 Jess in Action web site 173 Jess prompt and I O routers 319 Jess values sending to Java 309 jess awt ActionListener 208 jess awt TextAreaWriter 320 jess awt TextReader 320 jess Console 35 289 320 implementation 321 jess ConsoleApplet 289 jess Context 242 244 308 416 jess Deftemplate 240 311 jess Fact 130 222 240 311 416 constructing ordered facts in Java 313 constructing unordered from Java 345 constructing unordered in Java 312 ownership 312 jess FactIDValue 245 jess Funcall 240 245 315 417 jess jar 33 jess Jesp 417 jess JessEvent 418 jess JessException 241 316 jess JessListener 419 jess LongValue 245 jess Main 34 283 307 320 jess Node 145 jess PrettyPrinter 419 jess Rete 208 210 240 307 380 420 constructors 307 jess RU 43 139 241 422 jess Strategy 121 jess Token 130 423 jess Userfunction 235 jess Userpackage 248 253 jess Value 43 62 63 195 241 308 345 423 constructors 245 immutability 43 345 resolving 242 244 308 type 245 jess ValueVector 240 425 superclass of Fact 311 jess Variable 245 Jess gt prompt 77 JessEvent 418 JessException 316 catching 316 con
41. ing and the getLineNumber method points to a specific line in your script get ProgramText returns the actual snippet of Jess code that caused the error The code in this example provokes a JessException by trying to multiply two symbols and then demonstrates how to use some of these methods import jess public class CatchJessException public static void main String argv try Rete r new Rete r executeCommand 1 2 n 3 4 n a b catch JessException je System out print An error occurred at line 18 5 1 Working with JessException 317 je getLineNumber System out println which looks like je getProgramText System out println Message je getMessage Running this program produces the following An error occurred at line 3 which looks like ab Message Not a number a The most important thing to remember about working with JessExceptions is never to ignore them A truly remarkable number of people write code like this try Rete r new Rete r executeCommand 1 2 n 3 4 n a b catch JessException je NOTHING and then wonder why their program doesn t work Don t be one of them Always at least print the exception object itself to System out or to a log file JessExcep tion s toString method produces the by now familiar exception display you see from the Jess command prompt Printing the object je would give the
42. lt object gt Throws the given object which must be an instance of java lang Throwable or one of its subclasses If the object is a JessException throws it directly If the object is some other type of exception it is wrapped in a JessException before throwing The object s stack trace is filled in such that the exception appears to have been created by the throw function time Returns the number of seconds not milliseconds since 12 00 A M January 1 1970 try lt expression gt catch lt expression gt finally lt expression gt Works something like Java try with a few simplifications The biggest difference is that the catch clause can specify neither a type of exception nor a variable to receive the exception object All exceptions occurring in a try block are routed to the single catch block The variable ERROR is made to point to the exception object as an EXTERNAL_ADDRESS For example try open NoSuchFile txt r catch printout t call ERROR toString crlf Jess reported an error in routine open while executing open NoSuchFile txt r Message I O Exception An empty catch block is fine it signifies ignoring possible errors The expressions in the finally block if present are executed after all try and or catch code has executed immediately before the try function returns undefadvice lt lexeme gt ALL lt list gt Removes advice from the named function s see defadvice ALL ind
43. my fault I thank Andrew Grothe Bob Tre lease David Young Jeff Wang John Crabtree John Mitchell Mark Watson Michael J Smith Roedy Green Said Tabet Ted Neward and Daniel Selman and of course Bob Orchard for reading and commenting on the manuscript ACKNOWLEDGMENTS XXV I deeply appreciate the support I ve received from my management at Sandia both for encouraging Jess s development over the years and for permission to write this book in my copious free time Thanks to Paul Nielan Ken Washington Jim Costa Len Napolitano and Mim John I m also deeply indebted to my innova tive business partner Craig Smith who handles Jess licensing with aplomb Finally I want to thank my family for their encouragement and support Every year with a preschooler is an adventure and this last one has been no exception My wife Stacia deserves special thanks for picking up the slack when I was busy writing And to my daughter Danielle by the time this sentence is printed I bet you will be able to read it I love you both this book is for you about this book This book was originally conceived in August 2001 As I write these words now in May 2003 I feel like I ve stayed quite close to the original concept for the book Then as now despite the still growing prominence of rule based systems in nearly every field of software development the few available books on the topic were heavily theoretical and lacking in real world
44. number of feet gt See Go Eguchi and Laurence Leff XML Rule Editor for Java Expert Systems Proceedings of the In ternational Conference on Artificial Intelligence and Law ICAIL 2002 66 71 Laurence Leff Auto mated Reasoning with Legal XML Documents in ICAIL 2001 215 216 and Laurence Leff Rule Processing in the Legal XML Context in ICAIL 2001 27 30 20 4 Representing Jess rules in XML 367 lt hair gt yes lt hair gt lt color gt black and white lt color gt lt animal gt Code to work with this kind of XML is shorter and more efficient than code to work with the more general representation we looked at earlier In a specialized domain this kind of representation allows you to develop dedicated tools more quickly and easily The downside of course is the loss of generality In the next section you ll develop an XML format for Jess rules that uses the first of these two approaches You shouldn t interpret this to mean that I think the first approach is superior Each approach has advantages and disadvantages and you should select an approach after considering the alternatives carefully Representing Jess rules in XML In this section you re going to develop an XML representation for rules Rather than being motivated by theoretical goals this representation has the modest goal of being useful as a file format for editing and storing Jess rules You ll develop an XML DTD Data Type Definit
45. represent logical and and pipes represent logical or For example this pattern matches any client that hails from Boston or Hartford client city Boston Hartford And this one again matches any client not from Bangor and in addition remem bers the contents of city in the variable c client city c amp Bangor When you use several connective constraints together in a single expression you should pay attention to operator precedence or the way Jess groups the constraints together as it evaluates the expression The connective constraint has the high est precedence followed by amp and in that order always applies to the single constraint immediately following it so the following redundant pattern matches all clients that are not from Bangor and are from Portland client city Bangor amp Portland This pattern does not mean all clients that are from neither Bangor nor Port land which would be written client city Bangor amp Portland There are no grouping symbols that you can use with constraints you can t use parentheses to change their precedence If you can t express what you want using connective constraints you can do it instead using predicate constraints as described in the next section Constraining matches with predicate functions Literal constraints variables and connectives suffice for many situations but there are some things they can t express Perhaps you want
46. servlet container is to a servlet what an operating system is to a device driver the framework to which servlets are attached The container provides essential ser vices to the servlet such as pipes to and from the network methods to retrieve query data and configuration information and a mechanism for communication with other servlets in the same container Modern servlet containers are powerful things The J2EE environment of which they are a large part and its standard ized procedures for configuring installing and managing applications are a far cry from the early days of ad hoc web applications Developing and deploying web applications is easier than it has ever been Tomcat is a servlet engine from the Apache project http jakarta apache org tomcat index html Tomcat is the reference implementation for the Java Servlet and JavaServer Pages specifications and it s one component of JavaSoft s imple mentation of the Java 2 Enterprise Edition J2EE You can download and use Tomcat free of charge under the terms of the Apache Software License All the APIs and techniques you will use are general but I ll describe them in terms of Tomcat I ll assume for the sake of clarity that you re using a Linux computer with a standalone Tomcat installation The path to the root of the Tomcat installa tion is in an environment variable named Tomcat If your computer is running Windows instead of Linux you ll need to adjust the paths a
47. so when it is displayed to the console the embedded quotation marks are escaped with a backslash character It is important to note that all the work of pattern matching comparing the LHSs of rules to a given fact is done while that fact is being asserted Because initial fact is asserted by the reset function null rule is activated whenever the reset function is called and that s what happens here The function call watch rules tells Jess to print a message whenever a rule is fired A rule is said to be fired when the actions on its RHS are executed The run function tells Jess to start firing rules no rules will fire except during a call to run Jess s rule engine then fires the rules on the agenda one at a time until the agenda is empty run returns the number of rules fired so 1 is printed in the pre vious example Now let s look at a more complex rule Jess gt defrule change baby if wet If baby is wet change its diaper wet lt baby is wet gt change baby retract wet TRUE This rule again has two parts separated by gt The LHS consists of the pattern baby is wet The RHS consists of two function calls to change baby and retract Note that the definition of change baby isn t shown here Although you might at first find it hard to tell due to the Lisp like syntax the LHS of a rule con sists of patterns that are used to match facts in the working memory while the RHS contains function c
48. the code to do these conversions in a single class named State Part of the straightforward State class is shown in listing 14 4 Listing 14 4 Converting between the Hardware states and convenient symbolic names package control public class State public static final String OPEN open CLOSED closed OFF off HEATING heating COOLING cooling public static String vent boolean val return val OPEN CLOSED public static boolean vent String val if val equals OPEN return true else if val equals CLOSED return false else throw new IllegalArgumentException val Analagous heatpump methods not shown By collecting these conversions together in a single class and defining the symbols as constants in one place you may avoid a lot of debugging The Vent bean The other JavaBeans you need to write are a little different from the Thermometer class Whereas the Hardware interface only has a method for reading the temper 234 CHAPTER 14 The reality connection ature there are methods for both getting and setting the heat pump and vent states The corresponding HeatPump and Vent Beans therefore need both setting and getting methods Calling vent setState for example should fire a Proper tyChangeEvent Of course there should also be a background thread watching for changes from the outside You can reuse the BeanSupport base class to help implement the event support The Vent
49. they are simply ignored Note that Jess comments can nest a semicolon can appear in a comment without trouble Many Jess programmers use multiple semicolons on a line for visual emphasis RE Ee ror Gene ane LEG ne a oe T nea A Rules added 2 3 2003 Ut a 0 0 0 2 Although you ll see this usage often in Jess code it has no special meaning Whitespace symbols numbers strings and comments are the fundamental elements of Jess syntax Now you re ready to learn how to put these elements together into code Adding some structure 45 4 2 Adding some structure 4 2 1 Tokens by themselves aren t terribly interesting any more than isolated words have much to say Tokens become meaningful when they are put together into code In the following sections you ll see how to put together sentences in the Jess language Lists The basic unit of structure in Jess code is the list A list is a group of tokens that includes an enclosing set of parentheses and zero or more tokens or other lists The following are all valid lists 3 2 a b c Hello World O 1 2 3 deftemplate foo slot bar Lists are a little like arrays in Java and other languages In Jess they re the central way of structuring both code and data The first element of a list is called the list s head and in many contexts it is special For example Jess function calls are lists where the head is the name of a function and rules are written
50. type number then return numberp answer else return gt str length answer 0 The second parameter type to this function can be yes no number or anything else If it is yes no the function returns FALSE unless answer is yes or no If 168 10 6 2 CHAPTER 10 Designing the application type is number then the function returns true only if answer is a number using the built in numberp function to test for this condition If type is any thing else is of type returns TRUE unless answer is the empty string Now it is easy to rewrite ask user to use is of type for error checking While you re at it you can use the new type parameter to enhance the prompt by add ing a hint about the possible answers deffunction ask user question type Ask a question and return the answer bind answer while not is of type answer type do printout t question if eq type yes no then printout t yes or no bind answer read return answer Again you should test these new functions Jess gt is of type yes yes no TRUE Jess gt is of type no yes no TRUE Jess gt is of type maybe yes no FALSE Jess gt is of type number abc FALSE Jess gt is of type number 123 TRUE Jess gt ask user What is the answer yes no What is the answer yes or no 42 What is the answer yes or no yes yes This time when I entered 42 as the answer the function reject
51. wet change baby if wet 1 1 1 t TRUE Jess gt assert baby is wet gt f 1 MAIN baby is wet gt Activation MAIN change baby if wet f 1 lt Fact 1 gt Jess gt run FIRE 1 MAIN change baby if wet f 1 Baby is now dry lt f 1 MAIN baby is wet 1 The watch all command in listing 7 1 tells Jess to print diagnostics for everything important that happens while this program runs Many of the diagnostics in the listing are interesting You first see how issuing the reset command again asserts the fact initial fact Although this rule won t need the initial fact in most programs the initial fact will be needed by many rules so you should always issue a reset command at some point before running a program When the rule is entered at the Jess prompt you see the line 1 1 1 t This result tells you something about Jess interprets the rule internally see chapter 8 for more information When the fact baby is wet is asserted you see the diagnostic Activation MAIN change baby if wet 1 This means Jess has 100 7 1 1 7 2 CHAPTER 7 Writing rules in Jess noticed that all the LHS conditions of the rule change baby if wet are met by the given list of facts here the single fact 1 and an activation record has been created Note how the activation record associates the specific fact with the rule this action will be important later Again the rule doesn t fire until you issue the run comm
52. while lt Boolean expression gt do lt expression gt The first argument in the while function must be a Boolean expression an expres sion that evaluates to TRUE or FALSE The while function evaluates the Boolean expression and if the expression is not FALSE all other expressions in the body of the while except the optional symbol do are evaluated The Boolean expression is then reevaluated and the cycle repeats until the expression becomes FALSE The value of the last expression evaluated is the return value invariably it is FALSE Here is an example of using while to add the numbers from 1 to 10 Jess gt bind i 1 1 Jess gt bind sum 0 0 Jess gt while lt i 10 do bind sum sum i bind i i 1 FALSE Jess gt sum 55 Note that if the Boolean expression is FALSE the first time it is evaluated the body of the while is never executed The while loop implicitly makes a decision about what code to execute each time it evaluates the Boolean expression The if then else function lets you make this kind of choice explicitly if then else You can use the if function to choose what code to execute next The syntax of Jess s if function is similar to the if statement in Java if lt Boolean expression gt then lt expression gt else lt expression gt The first argument to if is a Boolean expression If the value of the expression is not FALSE the if function evaluates the group of exp
53. you can use the compressed form new Funcall watch engine arg all execute context Some Jess functions can fail due to either programmer error or a runtime condi tion In the next section you ll learn how to deal with errors during execution Working with JessException Jess s Java API reports errors by throwing instances of jess JessException Therefore whenever you work with Jess in Java you need to catch this exception Working with JessExceptions can be a little tricky as exception classes go but using them correctly can save you a lot of head scratching when you re debugging your programs Here are some useful methods in jess JessException getContext Returns a Jess stack trace getLineNumber Returns the offending line number in Jess language code getNextException Returns a nested exception getProgramText Returns the offending Jess code itself getRoutine Returns the name of the Jess function or Java method An instance of JessException can of course provide an error message and a Java stack trace like all Java Throwables If the error comes from executing code in the Jess language the JessException can provide additional information about what went wrong The getContext method for example returns a kind of Jess lan guage stack trace showing what functions and constructs were executing when an error occurred The getRoutine method tells you what Jess function was execut
54. 1989 XCON included 17 500 rules and knew about 31 000 hardware components The estimated savings to DEC at the time was 40 million annually due to increased accuracy reduced testing costs and higher customer satisfaction compared to configuration by human workers Such systems have become common not only in manufacturing but in the mail order and Internet sales industry where rule based systems help to recommend related products 1 D O Connor and V Barker Expert Systems for Configuration at Digital XCON and Beyond Com munications of the ACM 32 no 3 1989 1 2 3 Some real world examples 11 optimize packaging and perform other routine order configuration tasks Part VI of this book describes a series of web enabled systems for order configuration built around the Jess rule engine using Java servlets and JavaServer pages Implementing business rules Corporations invariably define policies and strategies that specify how the busi ness should respond to events ranging from individual sales to hostile takeover attempts A business ruleis a policy or strategy written in executable form such that a computer can follow it Here are two simple examples of business rules govern ing common situations IF employee s length of service gt 1 year AND employee category is regular employee AND employee contributes to 401k plan THEN employee is vested in 401k plan END IF customer order is more than ten units A
55. 2 2 Calling static methods Static or class methods in Java are those methods that can be called without refer ence to a specific object In both Java and Jess code you can use just the name of a Java class to invoke any of its static methods One well known example is the java lang Thread sleep method Jess gt call Thread sleep 1000 pause for one second Jess gt You don t need to use the fully qualified name java lang Thread because the classes in Java s java lang package are implicitly imported in Jess When you call a static method you must include the call function name as shown in the example therefore the most common use of ca11 is to invoke static methods Jess includes other functions analogous to call to help you invoke other categories of methods as you ll see in the next section 5 2 3 Calling set and get methods Special Java objects called JavaBeans play an important role in Jess as you ll see in chapter 6 Therefore Jess includes many tools for working with them One of these tools is a pair of methods to simplify accessing their data Methods that look like the following are fairly common in most object oriented languages public String getName return name ne void setName String n name n They are often called accessors and mutators or getters and setters They are very common in Java and form an important part of the JavaBeans specification We ll talk more about JavaBeans in chap
56. 20 Jess XML and the enterprise Unfortunately restricting your development to that common core can be frus trating because rule engines vary quite a bit in their capabilities Furthermore standard rule representations may not be capable of expressing everything that any one specific engine can Still the benefits of interoperability often outweigh these disadvantages Editing and other processing In each of the applications you ve developed in this book the programmer has written all the rules after consultation with experts Some applications though are simple enough that nontechnical people can add rules at least they could if they understood how to program in a rule language In general you won t be able to teach programming to the folks in marketing and you shouldn t have to As a programmer there s no reason why you can t put together a rule editor a graphical interface that allows nontechnical people to add and modify the rules by pointing and clicking instead of programming Some rule engines come with an integrated rule editor or a toolkit for develop ing rule editors Sometimes these editors only let you create rules in the native language of the rule engine not in vendor independent XML If you want to keep interoperability you may need to develop an editor yourself The good news is that XML makes this development fairly painless An editor that works with XML doesn t need to worry about parsing existin
57. 230 232 design problem in 231 pseudocode 153 162 public classes 63 public constructors 63 Q queries 96 301 query trigger fact 130 R radiator 257 random 406 rate of change 276 RDBMS 351 read 167 195 406 readline 319 320 406 Recommendations Agent xxvii 288 291 292 296 324 329 331 relation 128 relational database 75 128 351 353 remote interface 379 380 remote method invocation 379 removePropertyChangeListener 92 230 repaint 207 replace 406 reporting tools 363 RequestDispatcher 334 requirements 151 184 requirements documents 153 requirements engineering 151 reset 48 49 77 79 80 97 99 114 170 205 210 258 311 406 implementation 139 resolveValue 242 respect 154 rest 50 58 406 Rete pronunciation 136 Rete algorithm 39 96 134 136 brief description 136 memory usage 141 optimization 139 performance 141 Rete network 21 136 143 viewing 144 Rete utilities 43 422 retract 78 94 100 107 139 313 314 406 efficient from Java 314 retract string 407 return 56 127 128 407 from a rule 128 from RHS of rule 169 return value constraint 260 297 implementation 107 return values type conversions 64 RHS See right hand side RI 386 right memory 138 right to left adapter 144 right hand side 17 42 97 RMI 379 class server 383 security manager 382 rmic 379 382 rmiregistry 382 robots 14 round 407 router 319 RU ADD 139 RU ATOM 64 251 RU
58. 256 control flow 50 control structures writing 244 control systems 222 conversions 241 cookies 339 COOL 33 CORBA 241 count query results 131 396 Craig Smith xxv create 5 50 214 396 crisp logical states 264 crlf 43 167 current module 123 customers 151 155 Cygwin 169 428 D DAML 364 Danielle xxv DARPA 364 data structures 166 designing 24 database triggers 128 databases 302 relational 75 128 351 353 Dean Mike 364 debugging 295 web applications 341 declarative programming 15 369 declare auto focus 127 191 303 node index hash 142 salience 122 variables 130 302 defadvice 58 215 396 default 83 default constructor 63 default slot values 83 163 default dynamic 84 defclass 87 90 118 232 258 315 396 differences from CLIPS 32 deffacts 80 94 169 174 200 deffunction 56 135 208 302 reading 60 defglobal 48 205 258 260 283 definstance 88 90 91 93 94 222 232 258 315 397 differences from CLIPS 32 defmodule 123 166 303 differences from CLIPS 32 defquery 128 168 301 305 339 internal variables 302 defrule 96 361 deftemplate 5 82 135 153 162 294 302 __clear 87 _ fact 87 __not_or_test_CE 87 designing 162 generated 89 implied 86 redefining 83 sharing between modules 166 169 defuzzification 268 delete 397 DeMorgan s rules 110 dependents 178 care expenses 176 179 number of 175 deployment descriptor 326 depth 409 conflict resolution strategy 201 desk
59. 327 358 360 ATTLIST 367 ELEMENT 367 XML editors 361 xsl template 370 XSLT 361 363 369 transforming rule representa tions 365 Z ZFuzzySet 267 More Java Titles from Manning Aspect in Action Ramnivas Laddad 1 930110 93 6 Softbound 512 pages 44 95 July 2003 A practical guide to Aspect Oriented Pro gramming AOP and AspectJ The reusable code examples that are provided will enable quick implementation of functionality in your system For easy ordering call 1 800 247 6553 or fax 1 419 281 6883 For online browsing secure ordering special promotions and fast delivery visit www manning com More Java Titles from Manning JUnit in Action Vincent Massol with Ted Husted ISBN 1 930110 99 5 Softbound 360 pages 44 95 September 2003 A guide to unit testing Java applications includ ing J2EE applications using the JUnit frame work and its extensions this book is about solving real world problems such as unit test ing legacy applications writing real tests for real objects automating tests testing in isolation and unit testing J2EE and database applications Using a sample driven approach various unit testing strategies are covered Vincent Massol with Ted Bed For easy ordering call 1 800 247 6553 or fax 1 419 281 6883 For online browsing secure ordering special promotions and fast delivery visit www manning com More Java Titles from Manning Eclipse in Action David
60. ATTLIST group name CDATA REQUIRED gt lt ELEMENT pattern slot gt lt ATTLIST pattern name CDATA REQUIRED binding CDATA gt Next slot elements have name attributes and contain zero or more variable constant and function call elements lt ELEMENT slot variable constant function call gt lt ATTLIST slot name CDATA REQUIRED gt Note that this restricts the kind of Jess constructs you can use You have to use named slots so you can only use unordered facts Furthermore you can t use the or connective Although this restricts the syntax you can use it doesn t restrict the patterns that can be expressed For instance instead of writing a pattern like ball color red blue you can write instead the following equivalent ball color color amp or eq color red eq color blue Remember that the Jess code is generated automatically from the XML so this restriction really doesn t matter Finally a variable only has a name whereas a function call contains a head followed by zero or more constants variables or other function calls The head and constant elements both contain plain text lt ELEMENT variable EMPTY gt lt ATTLIST variable name CDATA REQUIRED gt lt ELEMENT function call head constant variable function call gt lt ELEMENT head PCDATA gt lt ELEMENT constant PCDATA gt
61. Any input output streams and event listeners are not saved during the serialization process If the filename contains special charac ters it must be passed as a double quoted string build lt string gt Evaluates a string as though it were entered at the command prompt Attempts to parse and evaluate a single expression from the given string as Jess code If suc cessful returns the return value of the expression This function is typically used to define rules or other constructs from Jess code at runtime For instance build defrule my rule foo gt bar Note that the string must consist of a single expression Jess functions 395 call lt external address gt lt lexeme gt lt lexeme gt lt expression gt Calls a Java method on the given object or a static method of the class named by the first argument The second argument is the name of the method and subse quent arguments are passed to the method Arguments are promoted and over loaded methods are selected precisely as for new The return value is converted to a suitable Jess value before being returned Array return values are converted to Jess lists The functor call may be omitted if the method being called is nonstatic The following two method calls are equivalent i These are both legal and equivalent call vector addElement new java lang String Foo vector addElement new java lang String Foo call on engine lt external ad
62. Gallardo Ed Burnette and Robert McGovern ISBN 1 930110 96 0 Softbound 416 pages 44 95 May 2003 This book provides a thorough guide to using David Gallardo d Burette Robert McGovern Eclipse features and plugins effectively in the context of real world Java development Inside you ll learn how to use plugin tools for using Bii Eclipse in a team environment including us ing Ant and CVS for source control For easy ordering call 1 800 247 6553 or fax 1 419 281 6883 For online browsing secure ordering special promotions and fast delivery visit www manning com More Java Titles from Manning Bitter EJB tent Bruce Tate Mike Clark Bob Lee Bob Lee nilaan and Patrick Linskey ISBN 1 930110 95 2 i SMBITTER Softbound 440 pages 44 95 June 2003 y L J B In Bitter EJB Bruce Tate and his co authors con tinue the entertaining and engaging writing style established in Bruce s best selling first book Bitter Java as they seek to explore and ie explain antipattern themes This more ad vanced book dives deeper into antipatterns or common traps within the context of EJB technology For easy ordering call 1 800 247 6553 or fax 1 419 281 6883 For online browsing secure ordering special promotions and fast delivery visit www manning com More Java Titles from Manning EJB Cookbook Benjamin G Sullins and Mark B Whipple 1 930110 94 4 Softbound 352 pages 42
63. HashMap gt This example uses the wildcard character to mean import all the classes in this package but you can also import one class at a time by using the fully quali fied name Just as in Java the entire java lang package is implicitly imported so you can create Integer and String objects without importing that package explicitly So far you ve used HashMap s default constructor Of course you can create objects using a class s other constructors as well HashMap has a constructor that takes a Java int and a Java float as arguments If you invoke this constructor and pass normal Jess numbers Jess will make it work Jess gt bind prices new HashMap 20 0 5 lt External Address java util HashMap gt Jess like any Java code can only invoke the public constructors of public classes in other packages If you want Jess to be able to construct instances of the classes you define be sure to make both the class and its constructors public When you call a Java method Jess converts the arguments from Jess data types to Java types as indicated in table 5 1 Generally when converting in this direc tion Jess has some idea of a target type The target type is the Java type that is needed in a given situation In the HashMap example the target types are int and float because those are the types of the formal parameters of the only HashMap constructor that takes two arguments When passing an argument to a Java con struct
64. Jess language the full power of Java is directly available This capability is shown p in figure 5 1 We ll discuss embedding Jess in larger Java sys e f a tems in great depth later in this book This dese chapter describes how you can create Java a ee awed a objects call their methods and otherwise inter act with Java without writing any Java code Figure 5 1 You can use Jess from Java Jess is therefore kind of scripting language and call Java methods from Jess for Java Aside from using Jess to build rule based systems you can also use Jess for exper imenting with Java APIs as a prototyping tool or even to build entire applica tions Want to find out what a particular API method does with a specific argument It s usually faster to start the Jess command prompt and type in a single line of Jess code than it would be to write compile and run a short Java program to do the same experiment Want to try different arrangements of a graphical interface You can create the windows buttons and other graphical components with a few lines of Jess code and then interactively assemble and rearrange them to your liking This is a great way to experiment with the sometimes surprising behavior of Java s layout managers the classes that arrange components inside an on screen container Creating Java objects In chapter 4 you learned about plain lists Although lists are useful they are not as powerful as the Map and
65. KEKE KKK KKK KKK KKK KKK KKK KEKKEKKE 1 Jess gt 11 2 Asking the user questions The first set of rules you ll write form the interview module As defined the pur pose of this module is to ask questions of the user to collect the inputs for the recommend module The interview module should ask the minimum number of questions and if a question becomes moot it shouldn t be asked at all For exam ple looking back at the knowledge collected in chapter 9 you can see that it s irrelevant how much interest income the user had if his total income is greater than 50 000 To handle this requirement you won t ask questions blindly The rule that asks each question will often need to match the existence or absence of some other fact so that a question won t be asked unless certain conditions are met All the questions you ll ask are represented by facts using the question tem plate defined in chapter 10 A deffacts containing all the potential questions is shown in listing 11 1 deffacts question data The questions the system can ask estion ident income type number text What was your annual income estion ident interest type yes no text Did you earn more than 400 in interest estion ident dependents type number text How many dependents live with you estion ident childcare type yes no IB oes pas Uae Ie gel eee ade text Did you estion ident text Did you estion ident Did you e
66. Many projects have requirements documents written descriptions of how a proposed system should behave Despite the best intentions such documents rarely capture the expectations for a system in enough detail to allow the system to be imple mented Often you can get the missing details only by talking to stakeholders the customers and potential users of the system People can also direct you to books web sites and other people who will help you learn about the problem domain These days it s common to suffer from information overload when you try to research a topic there are so many conflicting resources available that it s hard to know what information to believe The stakeholders in the system can tell you which resources they trust and which ones they don t 154 CHAPTER 9 Collecting the knowledge If you find conflicting information among otherwise trustworthy references during your research or hear conflicting statements during interviews don t be afraid to ask for clarification You ll need a strategy for resolving conflicts that hinge on matters of opinion Sometimes you can do this by picking a specific per son as the ultimate arbiter Other times especially on larger projects it s appropri ate to hold meetings to get the stakeholders to make decisions in a group setting Interviewing strategies Cultivating a good relationship with the people you interview is important This sounds so simple you might not think
67. Once you have defined a module it becomes the current module Jess gt get current module WORK Jess gt defmodule COMMUTE TRUE Jess gt get current module COMMUTE If you don t specify a module all deffacts templates and rules you define auto matically become part of the current module Jess gt deftemplate bus slot route number TRUE Jess gt defrule take the bus bus lt bus route number 76 have correct change gt get on bus TRUE Jess gt ppdefrule take the bus defrule COMMUTE take the bus Pbus lt COMMUTE bus route number 76 COMMUTE have correct change gt get on Pbus Note that the implied deftemplate have correct change was created in the COMMUTE module because that s where the rule was defined You can set the cur rent module explicitly using the set current module function Modules scope and name resolution A module defines a namespace for templates and rules This means two different mod ules can each contain a rule with a given name without conflicting for example Partitioning the rule base with defmodule 125 rules named MAIN initialize and COMMUTE initialize could be defined simultaneously and coexist in the same program Similarly the templates COM PUTER bus and COMMUTE bus could both be defined Obviously then Jess needs a way to decide which template the definition of a rule or query is referring to When Jess is compi
68. Set containers in Java s Collections API A plain list is a good choice for holding a grocery list but you really need something like a HashMap to hold a grocery price lookup table The HashMap would let you easily look up the price of any item in the table Jess s new function lets you create instances of Java classes For instance you can create a Java HashMap and store it in a variable with the following function call Jess gt bind prices new java util HashMap lt External Address java util HashMap gt Jess uses the type RU EXTERNAL_ADDRESS for the jess Value objects that hold arbitrary Java objects When you display an RU EXTERNAL_ADDRESS type you see a string that contains the name of the class You might expect instead that Jess Creating Java objects 63 would call the Java toString method on the contained object if Jess did this however the results could be confusing A java lang Integer object and a Jess value of type RU INTEGER act very differently but if Jess used toString to display EXTERNAL_ADDRESS objects they d both print as a number The fully qualified name java util HashMap requires a lot of typing and typ ing package names like java util can be error prone In Java you can avoid using package prefixes with the import keyword Jess has an import function you can use to do the same thing Jess gt import java util TRUE Jess gt bind prices new HashMap lt External Address java util
69. The set of browser requests generated by a single customer visiting TekMart com is called a session A session lasts longer than a single browser request but not as long as the whole lifetime of the web application The requests in a session are usually identified using session cookies small bits of data that are transmitted along with the HTTP header data and stored in the browser If a server sends a session cookie to a browser the browser is supposed to send it back with each request it makes in the same session The server can then use the cookies to provide some kind of context for the collection of requests in a session A servlet container should manage all of this transparently All a servlet needs to do is to request a Session object The servlet can then store information in the Session and the servlet can then use the Session information to identify the individual user The next two lines of Catalog doGet erase any previous session data and create a new HttpSession Then the login name and a new order number are installed in the session so the other servlets will be able to find them request getSession invalidate HttpSession session request getSession session setAttribute customerId customerId session setAttribute orderNumber String valueOf getNewOrderNumber The getNewOrderNumber method invokes a defquery on the shared Rete object private int getNewOrderNumber throws JessException ServletCo
70. This is easy to see on each cycle every book must be checked to see if it is overdue and the overdue ones must be checked against every borrower to find the right address On the average for many rules the worst case performance of this simple algorithm will be something like the B B expression extended to deal with any number of patterns and multiplied by the number of rules We could write the result RF where Ris the number of rules F is the total number of facts and Pis the average number of patterns per rule If P is 2 then the runtime will scale as the square of the number of facts doubling the number of facts will multiply the runtime by a factor of 4 136 8 3 8 3 1 CHAPTER 8 Under the hood how Jess works The Rete algorithm We can improve on the performance of this simple but inefficient pattern match ing algorithm by thinking about the source of its inefficiency The typical rule based system has a more or less fixed set of rules whereas the working memory changes continuously However it is an empirical fact that in most rule based sys tems much of the working memory is also fairly fixed over time Although new facts arrive and old ones are removed as the system runs the percentage of facts that change per unit time is generally fairly small The rules finding facts algorithm is therefore needlessly inefficient because most of the tests made on each cycle will have the same results as on the previous iter
71. This time you only need to look at the output of the system A complete run of the application is shown in listing 11 5 Listing 11 5 A typical session with the Tax Forms Advisor 2 java jess Main taxes clp Jess the Java Expert System Shell Copyright C 2001 E J Friedman Hill and the Sandia Corporation Jess Version 6 1 4 9 2003 Type your name and press Enter gt Bertram KKK KKK KKK KKK KKK KKK KK KKK KKK KKK KKK AK Hello Bertram Welcome to the tax forms advisor Please answer the questions and I will tell you what tax forms you may need to file KKK KKK K KKK KKK KKK KKK KKK KKK KKK KEKEKEKEEK What was your annual income 14000 Did you work in a home office yes or no yes Did you give more than 500 in property to charity yes or no no Will you be able to file on time yes or no yes Did you have losses from a theft or an accident yes or no no Did you move for job related reasons yes or no no Did you have unreimbursed employee expenses yes or no yes Did you have reimbursed employee expenses too yes or no no How many dependents live with you 7 Did you have dependent care expenses yes or no yes x Please take a copy of form 1040 Explanation Home office expenses Unreimbursed employee expenses x Please take a copy of form 1040A Explanation Income below threshold with dependents Please take a copy of form 2016EZ Explanation Unreimbursed employee expenses 184 11 7 CHAPTE
72. Toolkit from http www iit nrc ca IR_public fuzzy fuzzyJTool kit html There isn t space here to provide a full tutorial on the topic of fuzzy logic or FuzzyJess and the Fuzzy Toolkit For that you need to refer to other sources of information such as Bart Kosko Fuzzy Engineering Upper Saddle River NJ Prentice Hall 1997 Leferi Tsoukalas and Robert Uhrig Fuzzy and Neural Ap proaches in Engineering New York John Wiley amp Sons 1997 or the documentation provided with the Fuzzy Toolkit 266 CHAPTER 16 Writing the rules value defines the term young when describing a fuzzy concept or fuzzy variable such as age For some people age 25 is young for others age 35 is young The concept young has no clean boundary In most situations you could say that age 1 is definitely young and age 100 is definitely not young However age 35 has some possibility of being young its status usually depends on the context in which it is being considered In fact an age can have some possibility of being young and also some possibility of being old It belongs to more than one set at the same time with different degrees of membership in the sets The representation of this kind of information is handled by fuzzy set theory Unlike classical set theory which deals with objects whose membership to a set can be clearly described in fuzzy set theory membership of an element in a set can be partial that is an element belongs to a set wi
73. Vent Figure 14 1 Vent A six story building heated and cooled Dae by two heat pumps Each floor has Heat temperature sensors and an automated pump 2 Vent air vent This small building contains 14 a devices to connect to the control Vent software six sensors six vents and two heat pumps 224 CHAPTER 14 The reality connection 14 2 Defining the hardware interface 14 2 1 This system includes hardware and software components This is a book about software so perhaps we should get the hardware considerations out of the way first Assume that the building s control systems purchased from Acme HVAC Sys tems Inc include all the sensors and actuators you need In fact they re already connected to one master computer on which the control software will run Acme HVAC Systems has provided a library of functions you can call from C language code to turn the heat pumps on and off read the temperatures and operate the vents To make this concrete imagine that this is a complete list of the available library functions int getHeatPumpState int heatPump void setHeatPumpState int heatPump int state double getTemperature int sensor int getVentState int vent void setVentState int vent int state Note that these are C functions so on and off values like the state of each vent must be represented as type int because classic C doesn t have a boolean type like Java The first argument to every function
74. When you re through with your grocery shop ping you ll want to erase the list from working memory Removing facts with retract Ifa fact represents a true statement about the world then when the world changes that fact needs to be removed from the working memory You can remove individ ual facts from the working memory using the retract function Arguments for retract can be actual facts that is a jess Value object of type RU FACT which holds a reference to a jess Fact Java object or they can be numeric fact ids Let s retract the two facts you currently have in working memory one at a time Jess s working memory 79 Jess gt facts f 0 MAIN initial fact f 1 MAIN groceries milk eggs bread For a total of 2 facts Jess gt retract 1 TRUE Jess gt facts f 0 MAIN initial fact For a total of 1 facts Jess gt bind f fact id 0 Fetch Fact object sneer with fact id 0 Jess gt retract f TRUE Jess gt facts For a total of 0 facts This example retracts the grocery list fact using its fact id and the initial fact fact by using the Fact object directly you first have to look up the Fact object using the fact id function Both approaches work but if you already have a reference to a Fact object then using that reference is faster Using fact ids is more convenient when you re working interactively at the Jess gt prompt You retracted two facts one at a time It would have been poss
75. a database on demand In the example given here backward chaining is used to avoid computing the factorial of a number more than once the factorial of an integer is the product of every integer between 1 and the number itself inclusive for large numbers this value can be expensive to compute You use the deftemplate factorial to store computed factorials The fact factorial 5 125 signifies that the factorial of 5 is 125 Figure 7 1 shows how this example works The rule print factorial 10 won t fire unless a fact giving the factorial of 10 is present Because factorial has been registered for backward chaining with the do backward chaining function Jess automatically asserts the fact need factorial 10 nil This fact matches the need factorial pattern in the do factorial rule which fires and asserts the fact factorial 10 3628800 Finally this fact activates the print factorial 10 rule which fires and prints its output 118 CHAPTER 7 Writing rules in Jess defrule print factorial 10 factorial 10 r1 N need factorial 10 nil defrule do factorial Figure 7 1 need factorial 242 A pictorial representation of the assert factorial x r factorial example from the text To use backward chaining in Jess you must first declare that specific deftemplates are backward chaining reactive using the do backward chaining function Jess gt do backward chaining factorial TRUE If the template is unordered if
76. a simulator 248 Counting devices 249 Matching heat pumps and floors 250 Operating the hardware 251 Implementing a Userpackage 253 Testing 253 Summary 254 Writing the rules 255 16 1 16 2 16 3 16 4 16 5 16 6 16 7 16 8 The control algorithm 256 Knowledge engineering with truth tables 256 How heat pumps work 257 Using guard lines 257 Saving energy 258 Setting up 258 Controlling the heat pumps 259 Enough is enough 260 The moment of truth 261 Controlling the vents 262 The vent rules 262 Testing the whole system 263 Controlling with fuzzy rules 264 Fuzzy logic briefly 265 The Fuzzy HVAC Controller 269 Exploring the fuzzy controller 282 What s next 282 Summary 283 CONTENTS xvii Jess on the Web 287 17 1 17 2 17 3 17 4 17 5 17 6 17 7 17 8 17 9 Java architectures for the Web 288 Fat client applications 288 Thin client applications 290 A Jess application for the Web 292 Knowledge engineering 292 Designing data structures 293 Writing the rules 295 About testing 295 The recommend requirements rule 296 Recommending videos and DVDs 297 Conspicuous consumption 298 More media rules 299 Refining the recommendations 300 Some useful queries 301 Maintaining the order number 302 Cleaning up 303 Summary 304 Embedding Jess in Java applications 306 18 1 18 2 18 3 18 4 18 5 18 6 18 7 Getting started with the Jess library 307 The executeCommand
77. about Tom Tom isn t in positions one or four and he isn t wearing the hideous orange pants Tom isn t in position 1 or 4 and isn t wearing orange position of Tom amp n is p4 amp 1 amp 4 amp p1 amp p2 amp p3 pants color of Tom is c4 amp orange amp blue amp c1l amp C2 amp c3 There are only four positions but you know Tom s position is not 1 not 4 and not p1 p2 or 7p3 more constraints than there are possibilities This is actu ally a good sign it suggests that you have more than enough information to solve the puzzle You can place a similar number of constraints on Tom s fashion choices All that is left is to print out the set of variables p1 p4 and c1 c4 that solves the problem gt printout t Fred p1 cl crlf printout t Joe p2 t m c2 crlf printout t Bob p3 c3 crlf printout t Tom Ppa Pc exrlf cerl t The symbol gt separates the if part of the rule from the then part it specifies what the rule should do if all the requirements are satisfied Here it prints a table of results If you enter the code for the problem into Jess and then run it you get the answer directly The source for this problem is in the file rosencrantz clp and you can run it like this C Jess61 gt java classpath jess jar jess Main rosencrantz clp Fred 1 orange Joe 2 blue Bob 4 plaid Tom 3 red You see that exactly one set of variables satisfies t
78. all valid numbers 3 4 5 643 6 0E4 1D The Jess language has three numeric types RU INTEGER corresponding to Java int RU FLOAT corresponding to Java double and RU LONG like the Java type long The type of a numeric value is inferred when it is parsed The type RU LONG isn t used much in fact you can t type a long literal in Jess We ll talk about a function that turns a string into an RU LONG value after we ve discussed functions and strings 44 4 1 5 4 1 6 CHAPTER 4 Getting started with the Jess language Strings Character strings in Jess are denoted using double quotes Typical strings look like this n foo Hello World You can use backslashes to escape embedded quote symbols just like in Java Nonsense he said firmly Jess strings do not recognize any other Java style escape sequences In particular you cannot embed a newline in a string using n On the other hand real new lines are allowed inside double quoted strings they become part of the string This Jess string is equivalent to the Java string Hello nThere Hello There Strings are represented as jess Value objects of type RU STRING Comments You can add descriptive comments to your Jess code Programmer s comments in Jess begin with a semicolon and extend to the end of the line of text Here is an example of a comment This is a number 1 2345 Comments can appear anywhere in a Jess program
79. all you have to do is copy the source code into the top level of a web application s directory Tomcat recognizes the JSP extension and the first time the file is requested it is compiled into a servlet and executed Forwarding to a JSP A servlet and a JSP can collaborate easily Generally the servlet performs a compu tation stores the result in the HttpServletRequest object and then forwards the augmented request object to the JSP The JSP finds the data it needs in the Http ServletRequest and generates the HTML page For example suppose you wanted to let a servlet create the Rete object used in the previous section and then pass it to the JSP The relevant part of the servlet would look like this 1 For more information about designing applications with servlets and JSPs see Bruce A Tate Bitter Java Greenwich CT Manning 2002 334 CHAPTER 19 Deploying web based applications Rete engine new Rete request setAttribute engine engine ServletContext servletContext getServletContext RequestDispatcher dispatcher servletContext getRequestDispatcher hello jsp dispatcher forward request response return The JSP would be rewritten like this lt html gt lt page import jess gt lt jsp useBean id engine class jess Rete scope request gt lt head gt lt title gt Hello World lt title gt lt head gt lt body gt lt H1 gt lt engine addOutputRouter page out
80. attribute name Taken together this rule transforms the XML fragment lt variable name color gt into the Jess snippet color including that leading space Listing 20 2 XSLT script to transform the XML rule format into Jess rules lt xsl stylesheet xmlns xsl http www w3 org 1999 XSL Transform version 1 0 gt lt xsl output method text indent no gt lt xsl strip space elements gt lt Top level rule template gt lt xsl template match rule gt lt xsl text gt defrule lt xsl text gt lt xsl value of select name gt lt xsl text gt amp xA lt xsl text gt lt xsl if test priority t gt lt xsl text gt declare salience lt xsl text gt lt xsl value of select priority gt lt xsl text gt amp xA lt xsl text gt lt xsl if gt lt xsl apply templates select lhs gt lt xsl text gt amp gt lt xsl text gt lt xsl apply templates select rhs gt lt xsl text gt amp xA lt xsl text gt lt xsl template gt Representing Jess rules in XML lt Rule left hand sides gt lt xsl template match lhs gt lt xsl for each select group pattern gt lt xsl text gt lt xsl text gt lt xsl apply templates select gt lt xsl text gt amp xA lt xsl text gt lt xsl for each gt lt xsl template gt lt xsl template match group gt lt xsl text gt lt xsl text gt lt xsl value of select name gt lt xsl text
81. be created holding numeric values and you won t encounter this kind of error More templates A form has a code name like 1040A It also has a descriptive name like Federal income tax short form Therefore the form template might look like this deftemplate form slot name slot description Because the system will ask the user a number of questions you need a generic way to represent a question and its answer Although you don t know yet what the question asking mechanism will look like because you haven t written the code you can guess that the following two templates might be a good start 10 4 Templates you don t need 165 deftemplate question slot text slot type slot ident deftemplate answer slot ident slot text The question template ties a symbolic name for a question in the slot ident to the text of the question You ll use the working memory as a convenient database in which to look up the question text by identifier so that if two rules might need to ask the same question you won t have to duplicate the text You ll use the type slot to hold an indication of the expected category of answer numeric yes or no and so on The answer template ties the answer to a question A question and its corresponding answer will have the same symbolic value in their ident slots Once an answer for a given question exists you won t ask it again You ll develop the code that uses these templates in sec
82. bit but everything else should be the same Deploying the Hello serviet Deploying the Hello servlet is a five step process 1 Compile the servlet 2 Create a deployment descriptor 3 Create the directory structure for your web application 4 Copy the files into the right locations 5 Restart or start Tomcat J2EE and the Tomcat engine 327 Compile the servlet To compile the servlet you need to include the file TOMCAT common 1lib servlet jar in your class path The following command line should do the trick javac classpath TOMCAT common lib servlet jar Hello java Create a deployment descriptor A deployment descriptor is a short XML file named web xml that describes how the container should configure your servlet If you ve never worked with XML before the most important things to know are that first XML files are plain text just like Java source and second XML files have a picky syntax which is less for giving than HTML to which XML has a strong family resemblance Here s a basic web xml file lt xml version 1 0 encoding ISO 8859 1 gt lt DOCTYPE web app PUBLIC Sun Microsystems Inc DTD Web Application 2 3 EN http java sun com dtd web app_2_3 dtd gt lt web app gt lt servlet gt lt servlet name gt Hello lt servlet name gt lt servlet class gt Hello lt servlet class gt lt servlet gt lt web app gt Many other possible options can be included but you ll worr
83. chapter 7 Therefore the main problem Jess must solve is that of matching the rules in the rule base to the facts in working memory Jess has to perform the following steps in an infinite loop 1 Find all the rules that are satisfied by a set of facts in working memory 2 Form activation records out of these rule fact associations 3 Choose one activation record to execute 8 2 An inefficient solution The obvious implementation of pattern matching would be to keep a list of the rules and simply check each one s left hand side LHS in turn against the work ing memory forming a set of activation records for any that match After choosing one rule and executing it you could discard the set of activation records and start again You might call this the rules finding facts approach It is obviously not very efficient and doesn t scale well After every rule firing the system must recheck every fact against every rule Doubling the number of facts or the number of rules roughly halves the performance of the system It is difficult to analyze pattern matching algorithms like this one in the gen eral case because the actual performance is dependent on the makeup of work ing memory and on the exact nature of the rules For the example rule in the previous section though we can say that this naive algorithm will take time pro portional to the product B B on each cycle where B is the number of books and B is the number of borrowers
84. contents of working memory m facts Displays the contents of working memory m reset Initializes the working memory m retract Removes facts from working memory m watch Tells Jess to print diagnostics when interesting things happen Jess includes several functions that let you peer into working memory and see what s happening We ll cover these first so you ll be able to follow the action in later sections Jess s working memory 77 The watch function Many Jess programs are developed interactively by typing at the Jess gt prompt and watching the results It would therefore be useful to be able to see when Jess s working memory has changed You can use the watch function to tell Jess to print messages when various interesting things happen Depending on the arguments you pass to watch you can get Jess to report on several different kinds of events including changes to the working memory If you type the expression watch facts then you ll see a message whenever any facts are added or removed from then on The reset function initializes the working memory and creates the fact MAIN initial fact and you can see the fact being added to working memory if watch facts is in effect Jess gt watch facts TRUE Jess gt reset gt f 0 MAIN initial fact TRUE Jess gt unwatch facts TRUE The gt symbol given by Jess in a response means the fact is being added to work ing memory whereas a lt symbol is prin
85. create one application What is the J2EE The Java 2 Enterprise Edition J2EE http java sun com j2ee index html is a specification for a software infrastructure for enterprise applications written in Java It includes more than a dozen APIs each of which has its own acronym and its own detailed specification An implementation of the J2EE specification is called an application server and many commercial J2EE application servers are available JBoss http www jboss org is a very popular open source implemen tation An application server provides services and infrastructure common to many enterprise applications The J2EE provides a standard vendor neutral mech anism for deploying enterprise applications on an application server making it easy at least in theory to migrate an application to a new application server The term J2FE is also sometimes used to refer to Sun s own reference imple mentation of the J2EE specification This freely available reference implementa tion makes an excellent standards compliant vendor neutral development platform but it is not intended to be suitable for deploying enterprise applica tions Sun doesn t support it as a deployment platform and it doesn t have impor tant features like the ability to distribute an application over multiple processors Much of the J2EE reference implementation is based on components from the Apache Jakarta project an exemplary open source project that develops hig
86. credit for dependent care expenses if she files either Form 1040A or Form 1040 dependent care is special in this regard all the other special circumstances can only be claimed on Form 1040 The rule to recommend the dependent care expenses form is therefore as follows defrule form 2441 answer ident childcare text yes gt assert recommendation form 2441 explanation Child care expenses You know that if the childcare question has been asked and answered the user must have dependents and therefore isn t filing Form 1040EZ See the rule request childcare expenses in listing 11 3 For most of the other special circumstances you recommend Form 1040 along with the special form Here s a representative rule defrule form 4684 answer ident casualty text yes gt bind ex Losses due to casualty or theft assert recommendation form 4684 explanation ex recommendation form 1040 explanation ex You may notice a problem here Form 1040 may be recommended multiple times You need a rule to combine multiple recommendations into one The following rule does the trick The separate explanations are concatenated with a newline rep resented in Jess as a double quoted string containing a line break between them defrule combine recommendations r1 lt recommendation form f explanation e1 r2 lt recommendation form f explanation e2 amp e1 gt retract r2 modify r1
87. default handler You can do this using the try function As a first example let s call the function parseInt in the java lang Integer class which throws NumberFormatException if its argument can t be parsed as the string representa tion of an integer Jess gt deffunction parseInt string try bind i call Integer parseInt string printout t The answer is i crlf catch printout t Invalid argument crlf TRUE Jess gt parseInt 10 The answer is 10 Lowercase ell Jess gt parseInt 10 a uppercase oh Invalid argument The try function evaluates the expressions in the first block one at a time If one of those expressions throws an exception that block is abandoned and try begins evaluating expressions that follow the symbol catch if it appears You can also have a finally block following the catch block just like in Java The finally expressions are evaluated regardless of whether the try statements throw any exceptions A good use for a finally block is to close a file as in this Jess render ing of some typical Java file I O code Jess gt import java io TRUE Jess gt bind file nil Jess gt try bind file new BufferedReader new java io FileReader data txt while neq nil bind line file readLine printout t line crlf catch 72 CHAPTER 5 Scripting Java with Jess printout t Error processing file crlf finally if neq nil fi ile close Error
88. deftemplate y slot b TRUE Jess gt deftemplate z slot c TRUE Jess gt defrule example 1 x a v1 y b v1 gt TRUE Jess gt defrule example 2 x a v2 y b v2 z gt TRUE x y x y lt a gt LEFT 0 a RIGHT b LEFT 0 a RIGHT b example 1 Figure 8 1 An unoptimized Rete network for the example 2 2 two rules example 1 and example 2 138 CHAPTER 8 Under the hood how Jess works In this diagram each box represents a node A node s inputs are shown on the top and its outputs are on the bottom The diamond shaped nodes marked q are one input or pattern nodes The pattern nodes in this example test if the head of a fact is q Facts that pass this test are sent to the node s output others are ignored The trapezoidal nodes are two input or join nodes Each join node joins the results of matching the first n 1 patterns coming from upper left in the diagram with the nth pattern attached at upper right in the diagram Join nodes remem ber all facts or groups of facts that arrive on either of their two inputs The network is built so that the left input can receive groups of one or more facts the right input receives only single facts Every join node produces groups of two or more facts as at its output The arrivals from the two inputs are kept in separate memory areas traditionally called the alpha and beta memories We ll refer to them as the left and right me
89. delete any top level function calls to set vent state that were used for testing the last time around You should find that this set of rules controls the temperature of every floor to within 2 degrees of the set point While testing keep an eye out for heat pumps that are turned on and off quickly or that rapidly move between the heating and cooling state This set of rules should avoid both of these problems If you re using the version of the code downloaded from this book s web site then when the simulator starts the outdoor temperature is 90 degrees F and the set point is 70 so the heat pumps are in cooling mode Notice how the heat pump for the highest floors turns on first and runs the most this is a consequence of the heat rises property you built into the simulator The vents open and close fre quently to keep the temperatures of the three floors connected to each heat pump as close together as possible Once the system reaches equilibrium each heat pump spends some of its time turned off Watch the temperatures and make sure they all stay in the range between the set point and the upper guard line between 70 and 72 degrees 264 16 6 CHAPTER 16 Writing the rules Now set the outdoor temperature to 50 degrees As the temperature of each floor drifts down most of the heat pumps will shut off Eventually the lower floors will need to be heated and those heat pumps should move into heating mode The upper floors wil
90. develop a knowledge base to go with it This is the whole reason you re here The knowledge base is the collection of rules that make up a rule based system Rules can take actions based on the contents of working memory There are two main classes of rules in Jess forward chaining and backward chain ing rules Forward chaining rules are somewhat like if then statements in a pro cedural language and they re the most common and important kind of rule in Jess Backward chaining rules on the other hand don t have a clear analogy in procedural programming They are also similar to if then statements but a backward chaining rule actively tries to satisfy the conditions of its if part You can access working memory directly with queries You can design queries to search working memory to find specific facts and to explore their relationships Queries have a lot in common with rules if you can write one you know how to write the other You ll learn how to write and invoke queries in section 7 7 Forward chaining rules A forward chaining rule is something like an if then statement in a procedural language but it is not used in a procedural way Whereas if then statements are executed at a specific time and in a specific order according to how the programmer writes them a Jess rule s then part can be executed whenever the if part is satisfied This makes rules less obviously deterministic than a typical procedural
91. engine addUserfunction new CountDevices CountDevices N_HEATPUMPS engine addUserfunction new CountDevices CountDevices N_FLOORS engine addUserfunction new WhichPump engine addUserfunction new ReadHeatPump WriteHeat Pump engine addUserfunction new engine addUserfunction new ReadVent engine addUserfunction new WriteVent The Jess function call load package control HardwareFunctions now makes all the Userfunctions you ve written available from Jess Testing Testing these functions by hand using the simulator GUI is easy and actually kind of fun Load the functions using load package and then call init simulator 9 the GUI from figure 14 2 appears When you first see this window all the heat pumps are in the OFF state and all the vents are CLOSED The temperatures in the right hand column of the GUI are rising because when the simulator is turned on the outdoor temperature is a scorching 110 degrees F First check that you can count devices properly using the n heatpumps should return 3 and n floors should return 9 functions Make sure that which pump returns the correct values for each floor of the building should 254 15 7 CHAPTER 15 Extending the Jess language return 1 for arguments 1 3 2 for 4 6 and 3 for 7 9 You can and should write automated tests for these functions too perhaps using the testing framework in Appendix C Now turn on the heat p
92. engine executeCommand printout page Hello World from Jess crlf gt lt H1 gt lt body gt lt html gt The jsp useBean tag automatically extracts the Rete object from the HttpServ letRequest and assigns it to a variable of the correct type You can put the JSP forwarding machinery into a method so it can be called with one line of code You ll need such a method when you begin writing servlets it looks like this protected void dispatch HttpServletRequest request HttpServletResponse response String page throws IOException ServletException ServletContext servletContext getServletContext RequestDispatcher dispatcher servletContext getRequestDispatcher page dispatcher forward request response With dispatch the previous servlet body becomes the following Rete engine new Rete request setAttribute engine engine dispatch request response hello jsp return You ll see dispatch in all the servlets you ll write from now on The login screen 335 19 6 Application architecture take two 19 7 Let s reexamine the application architecture before you begin the implementa tion There are so many pieces now that it makes sense to summarize them in a table see table 19 1 Note that the order in which the files are listed is the order in which they ll execute during normal operation of the final application Table 19 1 The HTML files JSPs and servlets tha
93. even learned how to implement a simple sorting algorithm in a rule based language Note however that good design decisions don t happen spontaneously the Jess programmer will become more proficient over time just as in any discipline Experience and experimenta tion will help you develop good design skills In the next part of this book you ll research design and build another com plete system This one will be considerably more polished and will include a Java based graphical user interface You ll also see how backward chaining lets you eas ily write programs for problem diagnosis and troubleshooting Part 4 Writing a diagnostic application the PC Repair Assistant H developed one realistic application you re ready to move on to something bigger The application yov ll develop next includes both more knowl edge and more features It s a PC Repair Assistant with a graphical interface It s meant to guide a technician through the process of diagnosing computer hard ware problems In this part of the book we ll refine your understanding of engineering knowl edge by introducing a new tool flowcharting The program you ll write is much more sophisticated than the last one it is multithreaded and uses Jess s reflection capabilities to build a Swing based graphical interface directly from a Jess script Writing the PC Repair Assistant In this chapter you ll a Collect knowledge using flowcharts T
94. following Jess reported an error in routine Value numericValue while executing ab Message Not a number a type ATOM Program text a b at line 3 This is more information than the end user of your application probably needs to see but it s better than no information at all Nested exceptions The Jess function call can invoke any Java method If the invoked Java method throws an exception call creates a JessException object as a wrapper for the real exception and throws the wrapper instead Therefore when you catch a Jess Exception you may need to check it for a nested exception object using the get Cause method If getCause returns non nu11 the returned exception is almost always more interesting than the JessException itself import jess public class CheckNextException public static void main String argv try Rete r new Rete 318 18 5 2 CHAPTER 18 Embedding Jess in Java applications r executeCommand new java net URL foo bar catch JessException je System out println je getMessage System out println je getCause getMessage This code prints the following Constructor threw an exception unknown protocol foo Note that the toString method of JessException does not display nested excep tions so you need to check for and report them explicitly Rolling your own The call method of the Userfunction interface can throw JessException so if you re wr
95. for jess Context class In the next chapter you will write the Jess rules for controlling the Acme HVAC Systems devices The rules will pattern match the JavaBeans you developed in chapter 14 and call the functions you wrote in this chapter Then you ll assem ble and test the complete application Writing the rules In this chapter you ll a Write real time control rules Learn about fuzzy logic a Write a fuzzy version of the HVAC Controller 255 256 16 1 16 1 1 CHAPTER 16 Writing the rules You ve spent the last two chapters programming in Java developing the infrastructure for an HVAC Controller application In this chapter you ll assemble all the pieces and write the Jess rules to regulate the system In fact you ll write the rules twice First you ll write a set of rules that use simple logic and the Boolean vent states you programmed in chapter 15 they re Boolean because they re either open or closed If a given floor of the building gets too cold its heat pump will be turned on when it gets too warm its heat will be turned off This system will work much like a network of traditional mechanical thermostats and it will keep the tempera ture stable to within plus or minus one or two degrees Next you ll improve the control program using fuzzy logic and continuously variable vent controls The fuzzy rules will be able to adjust the vents to provide optimal air flow for each floor as a result
96. gt Jess gt facts f 0 MAIN initial fact f 1 MAIN person name Bob Smith age 34 gender Male f 2 MAIN person name nil age 30 gender Female For a total of 3 facts Sometimes nil is an acceptable default value but often it s not You can specify your own default value by using a slot qualifier Jess gt clear TRUE Jess gt deftemplate person People in actuarial database slot name default OCCUPANT slot age slot gender TRUE NOTE If you re following along you may have entered a previous definition for the person template You can t redefine a deftemplate until you use the clear function to completely erase Jess s internal state as this exam ple does 84 6 3 3 CHAPTER 6 Representing facts in Jess Notice that when you assert a person fact and don t specify a value for the name slot the default value OCCUPANT is used Jess gt assert person age 30 gender Female lt Fact 0 gt Jess gt facts f 0 MAIN person name OCCUPANT age 30 gender Female For a total of 1 facts What if the appropriate default value isn t constant but changes over time As an example a slot might be initialized to hold a timestamp indicating when the fact was asserted The default dynamic qualifier lets you accomplish this Jess evalu ates the given value each time a new fact is created using this template Usually you ll use a function call with default d
97. has a CLASSPATH variable defined you can sim ply add these two entries to the existing list Running Jess Jess is primarily intended as a library that can be embedded in other Java soft ware However when you re developing Jess code it s nice to have an interactive environment to work with so Jess comes complete with a simple command prompt To run Jess as a standalone command line application execute the class jess Main from the JAR file like this C gt java jess Main Jess the Java Expert System Shell Copyright C 2003 E J Friedman Hill and the Sandia Corporation Jess Version 6 1 4 9 2003 Jess gt Jess displays a welcome banner and then the Jess gt prompt When Jess displays this prompt it is waiting for you to enter a Jess program You can probably guess what s coming next You re going to write the Jess version of Hello World Hello World You can enter a Jess program directly at the command prompt and Jess will exe cute it immediately Jess gt printout t Hello World crlf Hello World Jess gt The Jess rule engine 35 You can also pass the name of a file that contains Jess code as an argument when you start jess Main Jess executes the code in the file and then exits C Jess61 gt java jess Main hello clp Hello World C Jess61 gt We will begin our study of the Jess language in the next chapter but if you re curious printout is a function that prints formatted output t te
98. if the expression is a function call When you call eval the string is parsed the expression is evaluated and the result is returned In this example a string is bound to a variable and then the eval func tion is used to evaluate the string and produce a result just as if the contents of the string had been typed at the command line Jess gt bind x 1 2 3 12 3 Jess gt eval x 6 The build function is a synonym for eval For historical reasons build is generally used to assemble rules and eval is used for function calls but they re identical Altogether Jess has almost 200 built in functions including these control structures If that s not enough and of course it never is you can define your own functions too as described in the next section Defining functions with deffunction Suppose that some time in the near future you find yourself in Dallas working on the AI module for a new martial arts video game you re using Jess of course Yow re writing many different rules that have to reason geometrically in particu lar many rules care about the distance between two combatants If the variables X1 Y1 and 21 hold the coordinates of one ninja and X2 Y2 and 22 the coordinates of another the Jess code to compute distance in three dimensions looks like this bind x X1 X2 bind y Y1 Y2 bind z Z1 Z2 bind distance sqrt x x y Py z z
99. is a good way to check for computational hot spots that might be slowing down your programs The matches function is easy to use Give it the name of a rule as an argument and it will show you information about each of the rule s join nodes in left to right order In the following example the rule matches demo s single join node has one token in its left memory and one in its right each memory gets its input from one pattern of the original rule Jess gt defrule matches demo a b gt TRUE Jess gt deffacts matches demo facts a b TRUE Jess gt reset TRUE Jess gt matches matches demo gt gt gt Node2 ntests 0 usecount 1 unique false Left Memory Token size 1 sortcode 1 tag ADD negcnt 0 facts MAIN a RightMemory Token size 1 sortcode 2 tag ADD negcnt 0 facts MAIN b TRUE Summary Jess uses the fast efficient Rete algorithm for pattern matching The strength of Rete is that it uses a set of memories to retain information about the success or fail ure of pattern matches during previous cycles The Rete algorithm involves building a network of pattern matching nodes Jess uses many different kinds of nodes to represent the many different kinds of pat tern matching activities There are also special nodes to handle some conditional elements like not and test as well as special behavior in some nodes to handle backward chaining This chapter concludes the introduction
100. is a need X fact Once this fact is asserted rules that match it can once again be processed in a forward chaining fashion To implement this functionality the join nodes that receive facts from a back ward chaining reactive template must act specially Each time a new left input token is received the number of successful pairings with right input tokens is counted If the count is zero a need X fact is generated The left input token rep resents the facts that match a rule up to but not including the pattern for the reactive template Once again you see that the left input token is crucial to the correct operation of this special type of join node and so Jess inserts initial fact patterns as necessary to provide them Exploring the Rete network in Jess The compiled Rete network is a complex and often large data structure Some times understanding it can mean the difference between an efficient program and a dreaded OutOfMemoryError In this section we ll look at some functions that will help you explore the Rete network watch compilations view and match The watch compilations command You can see the amount of node sharing in a Rete network by using Jess s watch compilations function Executing this function tells the rule compiler to print 144 8 7 2 CHAPTER 8 Under the hood how Jess works some diagnostics to the screen when each rule is compiled For each rule Jess prints a string of characters somethi
101. is done in Java code using classes from the FuzzyJ Toolkit definition of fuzzy variable Temperature with terms cold OK and hot FuzzyVariable temperature new FuzzyVariable Temperature 0 100 degrees C temperature addTerm cold new ZFuzzySet 8 15 temperature addTerm OK new triangleFuzzySet 8 15 25 temperature addTerm hot new SFuzzySet 15 25 definition of FuzzyValue for concept temperature is hot FuzzyValue tempHotFVal new FuzzyValue temperature hot fuzzifying the current crisp temperature various ways to do this here we represent an imprecison in the temperature sensor of 0 1 degree FuzzyValue currentTempFVal FuzzyValue temperature new PIFuzzySet 35 0 0 1 Without getting into too much detail this code creates a FuzzyVariable temper ature that provides the basis for building fuzzy concepts about temperature The domain for the variable is from 0 to 100 degrees C and the code defines three terms cold OK and hot that you can now use to represent specific temperature concepts The terms are defined using FuzzySets In this case cold is represented by a subclass of FuzzySet ZFuzzySet which defines a Z shaped fuzzy set like the one shown in figure 16 1 Once the FuzzyVariable and the terms that are to be used to describe con cepts about that variable are ready you can create FuzzyValues The example cre ates an object tempHot Ultimately
102. it helps you automate unit tests for Java soft ware There is no JessUnit per se as of this writing The framework TIl present here is the closest thing there is for now C 1 Architecture My Jess testing framework is based on Bourne shell scripts Therefore it s easy to use on UNIX like systems It can also be used on Windows systems by installing a UNIX emulation toolkit like the excellent free Cygwin environment Given the simplicity of the framework however you also could simply write your own using batch files The overall architecture of the test harness uses a top level tests directory con taining several template subdirectories and one additional subdirectory for each test To add a test you copy one of the template directories and edit the files in the copy Each directory including the top level directory contains a Makefile and a shell script named runtest sh The test directories contain additional files I ll describe these later Here s the directory structure for the framework before any tests have been added lA listing of some of these frameworks with links to more information is located at http xprogramming com software htm Cygwin is a porting layer that lets UNIX tools run on Windows The Cygwin home page is at http www cygwin com C 2 The Jess template 429 tests runtest sh runitest sh clean sh shelldefs mk JessTemplate Makefile README runtest sh runtest ref test clp
103. it needs to be said but it does Computer people have a culture all their own and it s different enough from mainstream culture that programmers can be perceived as rude If you re a computer pro grammer working as a knowledge engineer you may have to alter your accus tomed behavior when you re interviewing nonprogrammers Here are a few things to watch out for Speak their language It can be difficult for a programmer to remember that stacks loops shifts and pointers are not part of the everyday vocabulary of most nonprogrammers Don t use programming terms if you can avoid it You ll also want to avoid geek words like grok kludge and lossage which will only distance you from the interviewee Instead work hard to learn the technical jargon of the problem domain and use it properly Show respect No matter how trivial the domain may seem the interviewee knows more about it than you so don t look down on people just because they don t have the same education you do Your knowledge of program ming is not more important than their knowledge of inventory procedures Your time is not more valuable than theirs They re doing you a favor by talking to you so be grateful Be interested Make eye contact when you talk to the interviewee Ask follow up questions to show that you re listening Take notes so you don t ask the same question twice unless of course you didn t understand the answer the first time
104. long form Everyone needs it Form 1040A is the short form You can use it instead of Form 1040 if your taxable income is less than 50 000 You can t itemize deductions if you use this form but you can get a credit for child care expenses Form 1040EZ is the really short form You can use it instead of Form 1040A if you made less than 50 000 you have no dependents and you don t itemize deductions If you re married you and your spouse must file a joint return or you can t use this form Form 2441 lets you claim a credit for daycare expenses With Form 2016EZ you can deduct the unreimbursed part of any expenses you incurred for your employer primarily travel except commuting You can use this short form only if you weren t reimbursed for any expenses otherwise you have to use the long form Form 3903 gets you a deduction for unreimbursed moving expenses if you moved this year because of your job Form 4684 lets you recover some of your losses during the year the part that was not covered by insurance Many people use this form to deduct costs due to car accidents Collecting knowledge about tax forms 159 You fill out Form 4868 to get an extension for filing your taxes Note that you still have to pay your taxes on time you can pay an estimated amount with this form You need to file Form 8283 to get credit for donating more than 500 worth of property to charity You can file Form 8829 if you have a home offi
105. matches a line item in this customer s current order The pat tern after that matches the product represented by this line item but only if that product requires some other category of product This second pattern therefore identifies a case where you can make a recommendation line item order number currentOrder part number part product part number part name product requires category The fifth positive pattern matches some other product belonging to the category required by the first product The sixth negated pattern matches another prod uct in this category that costs more than the first Because it s negated it means there s no such product these two patterns together identify the most expensive product you could recommend product category category price price part number second part not product category category price p amp gt p price Now that you ve identified the product to recommend you can assert a recom mend fact for it gt order number currentOrder type requirement part number second part because product assert recommend This one long rule will be a workhorse in your system Writing the rules 297 17 5 3 Recommending videos and DVDs You ll recall that the folks in marketing are interested in selling customers some extra videotapes and DVDs There are four situations in which you should recom mend some kind of media when th
106. matching is driven by facts being asserted the matching computation hap pens during the assert definstance modify duplicate or reset function that creates the fact Because a not CE matches the absence of a fact when can it be evaluated The answer is that a not CE is evaluated only in these cases m When a fact matching it is asserted in which case the pattern match fails m When a fact matching it is removed in which case the pattern match succeeds m When the pattern immediately before the not on the rule s LHS is evaluated If a not CE is the first pattern on a rule s LHS the first pattern in an and group or the first pattern on a given branch of an or group the pattern initial fact is inserted before the not to become this important preceding pattern Therefore the initial fact created by the reset command is necessary for the proper func tioning of many not patterns For this reason it is especially important to issue a reset command before attempting to run the rule engine when working with not patterns The not CE can be used in arbitrary combination with the and and or CEs You can define complex logical structures this way For example suppose you want a rule to fire once and only once if for every car of a given color there exists a bus of the same color You could express that as follows Jess gt defrule forall example not and car color c not bus color c gt Decoding complex logical expr
107. name funcall gt lt xsl text gt lt xsl text gt lt xsl apply templates select gt lt xsl text gt lt xsl text gt lt xsl template gt lt xsl template match function call function call gt lt xsl text gt lt xsl text gt lt xsl call template name funcall gt lt xsl template gt lt Miscellaneous gt lt xsl template match variable gt lt xsl text gt lt xsl text gt lt xsl value of select name gt lt xsl template gt lt xsl template match constant gt lt xsl text gt lt xsl text gt lt xsl value of select gt lt xsl template gt lt xsl stylesheet gt _ Now that you ve decided on an XML representation and come up with a way to translate XML rules into Jess rules you need a way to create and modify rules in the XML format XML is a text based format so you could simply fire up your favorite text editor but there are definitely better alternatives Rule editors If rules are represented in XML documents then it follows that you can use XML editors to create rules Some XML editors are graphical in nature and you can edit a document by dragging appropriate elements into a tree If your XML for mat is described rigorously by a DTD then many editors can use it to restrict the documents you create to only well formed rule documents Although using an XML editor is certainly a possibility in general doing so is too complicated for nontechnical users An edit
108. name of the variable The third value is the new value for the vari able the same conversions are applied as for new and call See get member Jess functions 409 set multithreaded io TRUE FALSE Specifies whether Jess should use a separate thread to flush I O streams Turning this on can lead to a modest performance enhancement but may result in loss of output on program termination Returns the previous value of this property set node index hash lt integer gt Sets the default hashing key used in all Rete network join node memories defined after the function is called this function will not affect parts of the network already in existence at the time of the call A small value gives rise to memory efficient nodes a larger value uses more memory and increases performance up to a point If the created nodes will generally have to remember many partial matches large numbers will lead to faster performance the opposite may be true for nodes that will rarely hold more than one or two partial matches This func tion sets the default explicit declare statements can override this for individual rules The default value is 101 set reset globals TRUE FALSE nil Changes the current setting of the global variable reset behavior If this property is set to TRUE the default then the reset command reinitializes the values of global variables to their initial values if the initial value was a function call the function call
109. nodes for example test the head of a fact and therefore provide the entrance route into the network Node1TEQ nodes compare the value of a single slot to a constant value and Node1TEV1 nodes test whether two slots within a single fact contain the same value There are separate multislot variations of these types their names begin with Node1M Node1MTEQ nodes for example compare a single datum at a particular location in a multislot to a constant value Finally there are negated versions too Node1TNEV1 nodes test that two slots in a single fact do not contain the same value Besides the menagerie of nodes types listed here other types serve a structural role in the network Node1RTL nodes for example adapt an output that would normally lead to the right input of a join node so that it connects to the left input instead and Node1NONE nodes serve as stand ins for nodes that aren t really there specifically for the part of the pattern network that leads to the right input of a NodeTest The information that the matches or view functions display about individual nodes is generally fairly descriptive of each node s function 146 8 7 3 8 8 CHAPTER 8 Under the hood how Jess works The matches function The matches function lets you see the contents of the left and right memories of the join nodes used by any rule It can be an invaluable aid in determining why a particular rule isn t behaving the way you expect It also
110. of the execution of your system Most rule development languages offer explicit support for this kind of modularity and it s a good idea to use it when ever possible The driving robot s rules might be divided into separate modules devoted to starting the car parking the car city driving highway driving pass ing other cars and so on By breaking rules into small groups you can make a rule based program easier to write and to understand We ll first study writing rules for a real application an information kiosk in chapter 11 Iterative development Once you ve developed some rules you ll often find that you don t have all the information you need to write more When this happens you ll need to go back to the source and do some more knowledge engineering The development of a rule based system lends itself well to this sort of iterative procedure You can show the early incarnations of the system to the human experts if they exist and ask them for corroboration of the results You might have to change your tests if the experts disagree with what they are testing It s also worthwhile to have another knowledge engineer look over your work at this point Code reviews are amazingly effective at finding problems with soft ware before a release and they work for rule based software as well Whether you hold formal code reviews or just ask a friend for advice a second pair of eyes can really help to increase the quality of your wo
111. of the standards we ve just described because it will be tailored for your specific application With a bit of care it will be expandable to handle new requirements as your application evolves The best part about using a custom representation is that transforming XML from one representation to another can be easily accomplished using an XSLT script You can rest assured that it will be straightforward to translate your format into a future standard format or even into an improved version of your own for mat when the time comes Several of the commercial rule engines have their own XML based rule format but Jess does not You ll remedy this situation by developing one in the next section Strategies for representing rules in XML There are two broadly different strategies for representing rules in XML The most general strategy is the one taken by the DAML and RuleML projects In this strategy an XML element represents the concept of a pattern and the contents of the pattern are data inside the element As you ve seen the Jess pattern isa x mammal is represented in RuleML as lt fact gt lt atom gt lt _opr gt lt rel gt isa lt rel gt lt _opr gt lt var gt x lt var gt lt ind gt mammal lt ind gt lt atom gt lt fact gt 366 CHAPTER 20 Jess XML and the enterprise The interesting thing about this representation is that the data could change and the elements themselves could stay the same Fo
112. print String message Rete engine throws JessException engine executeCommand printout page message yy CrLPry jy This servlet is very similar to the first one The main difference is that HelloJess creates an instance of the Rete class installs the servlet s output Writer as a Jess router named page and then uses Jess s printout function via the helper method print to generate the HTML page There s nothing complex here but this example illustrates how simple it is to embed Jess in a Java application Deploying the Jess servlet The Hellovess servlet needs to have access to the Jess library to run so you must install Jess under Tomcat This is easy enough to do JAR files in the TOMCAT lib directory are automatically accessible to all installed web applications so you ll package Jess as a JAR file and put it there The binary Jess distribution already includes the JAR file jess jar If you have compiled your own copy of Jess from source you can build a JAR file like this on Linux jar cf jess jar find jess name class print jess clp Then copy jess jar to the TOMCAT 1ib directory In your Makefile or Ant build xml file and so on you may want to include a line that copies the jess jar file Modify the web xm1 file you used for the last servlet to account for the new servlet and application names install the application as HelloJess and restart Tomcat If you then visit the URL http
113. processing file le then You must have either a catch or a finally block or both in every try function call You can access the actual exception object that was thrown using the special vari able ERROR which is always defined for you in a catch block and is initialized by Jess to point to the caught exception Whereas in Java you can define multiple catch blocks in a single try differentiated by exception type you can have only one catch block in a try in Jess If you want to distinguish between exception types you can use the ERROR variable and the instanceof function which can tell you whether a given object belongs to a given Java class For example you may want to call the static method lookup in the java rmi Naming class This method is used to contact remote objects via the RMI protocol The lookup method can throw NotBoundException RemoteException and AccessException all from the java rmi package as well as java net MalformedURLException You can issue specific error messages for each of these cases the code might look like listing 5 1 DER eas import java rmi Lets you use S short class Jess gt import java net MalformedURLException names Service TRUE misspelled Jess gt try throws bind server NotBound call Naming lookup rmi snarf blat com Survice Exception printout t Connection established crlf catch if instanceof ERROR NotBoundException then Check exception print
114. program because Jess decides the order in which to fire the rules See section 8 3 s discussion of the Rete algorithm for an explanation of why this architecture can be many orders of magnitude faster than an equivalent set of traditional if then statements This section discusses the following functions and constructs defrule Defines a new rule podefrule Pretty prints a rule run Begins firing activated rules from the agenda undefrule Deletes a rule watch rules Prints a diagnostic when a rule fires watch activations Prints a diagnostic when a rule is activated All Jess rules are defined using the defrule construct The simplest possible rule looks like this Jess gt defrule null rule A rule that does nothing gt TRUE Forward chaining rules 97 The symbol null rule is the name of the rule A hyphen is often used to sepa rate words in a symbol Rules are uniquely identified by their name Ifa rule named my rule exists and you define another rule named my rule the first version is deleted There is also an undefrule function that can delete a rule by name The name is followed by an optional documentation string that describes the purpose of the rule The symbol gt an equals sign followed by a greater than sign separates the rule s left hand side LHS or if part from its right hand side RHS or then part The symbol gt can thus be read as then The previous rule has no conditions o
115. rules into XML and back 357 358 20 1 CHAPTER 20 Jess XML and the enterprise With this chapter we begin part 7 of this book This last part is different from the oth ers in that it s not structured around the development of a single application Instead we ll look at many of the pieces that might be used to build applications based on the Java 2 Enterprise Edition J2EE platform The J2EE is simply too big and too com plex and offers too many possibilities to be covered by one big example Therefore this part will be a kind of grab bag of individual topics It includes small programs and program fragments you can use to develop your own applica tions In this chapter we ll look at an overview of enterprise applications and cover representing rules in XML and software for editing rules In the next chap ter we ll look at using rules in Enterprise JavaBeans applications the javax rules rule engine API and more You can use the techniques we ll discuss to create new kinds of applications or develop enterprise ready versions of the applications we ve looked at so far These techniques are all aimed at providing practical solutions for the problem of deploying a rule based application in a business environment Enterprise applications The word enterprise has been the most durable information technology buzzword of the last decade Simply put an enterprise application is any large software application that your b
116. runtest sh script runs Jess using test clp as the input and captures the output in a file and then compares the result against test ref The file runtest ref contains the output of a successful run of runtest sh it s used by the top level scripts to determine whether the test passed The runtest sh script looks like this 430 C 3 APPENDIX C An automated testing framework shelldefs mk echo Running the test program S JAVA jess Main nologo test clp gt test out 2 gt amp 1 if diff ignore space change test out test ref gt dev null then files are the same echo Test succeeded else files are different echo Test failed Try echo diff ignore space change test out test ref fi The variable JAVA is defined in the top level file shelldefs mk which is read in at the beginning of this script All sorts of variable definitions can be collected in this file which helps you to write portable tests One important point about this script is that it uses the ignore space change option of GNU diff The diff program compares files and this option tells it to be lenient if the only changes involve extra spaces or blank lines This option makes it easier to write reference files you don t have to worry about get ting the spacing precisely right The file test clp looks like this deffunction test something printout t Testing erlf test something printout t Test done crlf exit
117. s working memory can contain ordered facts unordered facts and shadow facts each type of fact is useful in certain situations Unordered facts are general purpose facts whereas ordered facts are useful for small bits of information Shadow facts are used to connect a JavaBean in your Java application to Jess s working memory so that rules can react to things that happen outside of your Jess program You can write rules that operate on any or all of these working memory elements in the next chapter you ll see how All facts are created from a deftemplate which defines the slots a fact can have Sometimes you define this deftemplate yourself and sometimes Jess cre ates it for you The deftemplates in a rule based system are like the schema of a database they define a way of looking at the data relevant to the system Now that you know the Jess language and understand something about Jess s working memory you re ready to learn how to write rules In the next chapter you ll learn about writing rules and about pattern matching how to make rules react to the contents of working memory Writing rules in fess In this chapter you ll Learn to write rules Learn the difference between forward and backward chaining Learn how to partition your rules with modules Learn to probe working memory with queries 95 96 7 1 CHAPTER 7 Writing rules in Jess Now that you ve learned how to populate the working memory you can
118. simple text field to collect its answers for all question types This will give you a framework to build on later you ll replace this input field with a combo box and other more user friendly components First you ll build a panel containing a JTextField and a JButton labeled OK then you ll add it to the bot tom of the application window Listing 13 2 shows the code to do this the result looks like figure 13 3 I used get member to obtain the value of the constant SOUTH in the BorderLayout class The calls to validate and repaint are necessary whenever you modify the contents of a window that s already visible on the screen Listing 13 2 Adding an input area to the application window Add a JTextField and an OK button to our application window defglobal apanel new JPanel defglobal afield new JTextField 40 defglobal afield ok new JButton OK apanel add afield apanel add afield ok frame getContentPane add apanel get member BorderLayout SOUTH frame validate frame repaint Now attach an event handler deffunction read input EVENT 208 CHAPTER 13 Adding a graphical interface An event handler for the user input field bind text sym cat afield getText assert ask user input text bind handler new jess awt ActionListener read input engine afield addActionListener handler afield ok addActionList
119. stand for 359 CONTENTS xix 20 2 Rules and XML 360 Interoperability 361 Editing and other processing 362 Storage and retrieval 362 20 3 XML based rule representations 363 RuleML 363 DAML 364 Homegrown representations 365 Strategies for representing rules in XML 365 20 4 Representing Jess rules in XML 367 An example rule 368 Transforming the XML rules into Jess rules 369 20 5 Rule editors 372 20 6 Summary 374 Jess in the J2EE environment 375 21 1 A quick tour of EJB concepts 376 Kinds of EJBs 376 n EJB restrictions 377 Do you need to use EJBs 377 Accessing external resources from EJBs 378 21 2 An RMIl based rule server 379 The remote interfaces 379 Implementing the interfaces 380 Implementing a main method 382 Generating the stubs 382 A sample client 382 Final polishing 383 21 3 JSR 94 the javax rules API 383 Working with javax rules 384 The reference implementation 386 21 4 Summary 388 Jess functions 389 Abridged Java API for Jess 415 An automated testing framework 427 index 433 preface In 1994 I was working in the Scientific Computing department at Sandia National Laboratories in Livermore California We had an impressive for the time array of heterogeneous computing equipment workstations from Silicon Graphics and Sun Microsystems Intel PCs running Linux Macintoshes galore I was writing software agents that managed dynamically distributed computations across this networ
120. state heating number p Thermometer floor f amp eq p which pump f fuzzyReading t amp fuzzy match t cool gt assert fuzzy vent floor f fuzzy state new FuzzyValue ventFVar mostlyOpen There is just one thing left to do In order to set the vent position you need to define a rule that takes the combined output of all the rules and defuzzifies the result giving the required new setting for the vent Because the rule must wait until all the other fuzzy rules have fired to allow Jess to combine the outputs it is set at a lower priority salience than these rules The FuzzyValue method weightedAverageDefuzzify transforms the FuzzyValue into a crisp value this value is sent to the simulator to make the adjustment This final rule is coded as follows 276 CHAPTER 16 Writing the rules defrule set vent state declare salience 100 fv lt fuzzy vent floor f fuzzy state fs gt bind vent state fs weightedAverageDefuzzify set vent state f vent state retract fv When you run this example a sample batch file is provided to help and you will need the FuzzyJ Toolkit version 1 3 or later with FuzzyJess and Jess version 6 1 or later it will do a very good job of controlling the system With an outside tempera ture of 90 and a set point temperature of 70 the system quickly settles close to 70 degrees with each vent close to 0 5 open If the outside temperature is
121. structure 45 Lists 45 Calling functions 46 Variables 47 More about lists 49 4 3 Control flow 50 foreach 51 while 52 iffthen else 52 progn 53 apply 54 evaland build 54 4 4 Defining functions with deffunction 55 Late binding 57 CONTENTS xi 4 5 Fine tuning a function s behavior 58 Take my advice please 60 4 6 Summary 60 Scripting Java with Jess 61 5 1 Creating Java objects 62 5 2 Calling Java methods 64 Nesting function calls and a shortcut 65 Calling static methods 66 Calling set and get methods 66 Working with arrays 67 How Jess chooses among overloaded methods 68 5 3 Accessing Java member data 69 5 4 Working with exceptions 70 5 5 Summary 73 Representing facts in Jess 74 6 1 Jesss working memory 75 Manipulating the working memory 76 6 2 Just the facts maam 81 6 3 Unordered facts 82 The defiemplate construct 82 Default slot values 83 Multislots 84 Changing slot values with modify 85 Copying facts with duplicate 85 6 4 Ordered facts 86 6 5 Shadow facts 87 Jess and JavaBeans 88 JavaBeans have slots 88 An example JavaBean 89 Creating a defiemplate for DimmerSwitch 90 Putting a DimmerSwitch into working memory 90 Static vs dynamic shadow facts 91 Adding PropertyChangeListener support to DimmerSwitch 92 Shadow facts and working memory functions 94 6 6 Summary 94 CONTENTS Writing rules in Jess 95 7 1 Forward chaining rules 96 Patterns and shadow facts 100
122. supplying an object from your application as an argument to the Rete constructor you tell Jess how to find the other classes in your applica tion This will be important when you build a web application in the next chapter The executeCommand method Once your Java program creates a Rete object what s next One of the simplest and most powerful ways to manipulate Jess from Java is to use the executeCommand method executeCommand accepts a String argument and returns a jess Value see section 4 1 3 and chapter 15 to learn about this class The String is inter preted as an expression in the Jess language and the return value is the result of evaluating the expression So for example to add a fact to Jess s working memory and get access to the jess Fact object you might do the following import jess Rete engine new Rete Value v engine executeCommand assert color red Fact f v factValue engine getGlobalContext Section 15 2 2 describes value resolution how to get information out of a jess Value object To resolve a jess Value you need a jess Context object The method getGlobalContext in jess Rete is a convenient way to obtain one Commands executed via executeCommand may refer to Jess variables they are interpreted in this same global context Using executeCommand to interact with Jess has one other advantage It is thread safe Only a single call to executeCommand can be simultaneously execut ing on a giv
123. temp warm set vent mostly open rule will fire and a vent fact will be asserted with a mostlyOpen FuzzyValue scaled by the degree of matching of the tempera ture FuzzyValue In a complete system you normally have a number of rules that cover the range of decisions to be made Here for example the temperature might also match the FuzzyValue hot to some degree which would produce a vent fact with the FuzzyValue fullyOpen as an output again suitably scaled to represent the degree of matching These output facts would be combined into a single fuzzy output fact to represent the global decision of the system You can leave this combined output as a FuzzyValue but in general you ll go through a process of defuzzification of the output transforming the FuzzyValue to get a crisp value Doing so allows the system to take some real world action such as moving a vent by some real valued amount There are a number of ways to defuzz ify a FuzzySet One of the more common is to look at the area under the graph of the FuzzySet and find the center of gravity or center of mass Details of this and other types of defuzzification are found in the Fuzzy documentation Controlling with fuzzy rules 269 To wrap up and review here are the steps normally taken in a fuzzy control application 1 Collect the system inputs 2 Fuzzify the inputs make them appropriate fuzzy values usually repre senting the imprecision in the input value 3 Apply
124. test your application 14 1 The system 223 To complete the control rules cleanly you need to extend the Jess language by adding some custom functions written in Java this is the subject of chapter 15 We ll look at Jess s external function interface and learn something about how Jess s function call mechanism works under the hood Finally in chapter 16 you ll write the rules and the application will come together The system Imagine a tall building figure 14 1 The building is heated and cooled by a num ber of heat pumps A heat pump is both a heating and cooling device At any time it can be in one of three modes heating cooling or off In this building each heat pump services three adjacent floors Warm or cold air from the heat pumps enters the building through adjustable vents Each vent can be either open or closed and although there may be several vents on each floor they are wired in such a way that they must open and close in concert that is they act like a single vent You probably know that warm air rises and cold air sinks it s basic physics This adds a bit of nonlinearity to the system because each heat pump can affect not only the floors to which it directly sends heated or cooled air but the floors above and below it as well From temperature sensor To vent Vent Heat pump 1 aai Vent
125. that include an integrated rule engine generally make it available in the same way as an object you can retrieve from JNDI An RMI based rule server 379 21 2 An RMI based rule server 21 2 1 One way to use an external resource like a rule engine from an EJB environ ment is to wrap that resource in a Remote Method Invocation RMI server RMI is a J2EE API for communicating between separate Java applications either on one machine or over a network Most EJB environments use RMI for client server communications By letting the external resource run in its own separate process you can avoid problems due to EJB programming restrictions that the resource might violate RMI creates the illusion that an object in one Java application exists inside a second application so that that second application can call the methods of the remote object in the normal fashion The magic is accomplished by special classes called stubs and skeletons An instance of a stub class masquerades as a local object but its methods package up their arguments in a network request and forward the request to a remote object A skeleton class receives such a request unpacks the arguments and calls the method on the real object The return value if any is then packaged and returned to the stub s method which returns it as its own return value The end result is that for example an EJB could invoke methods on a rule engine as if it were running inside the J2EE s
126. the FuzzyValue is just a FuzzySet or at least it contains a FuzzySet but it also has a context the FuzzyVariable it is associated with You can perform operations on FuzzyValues For example the union and intersection of two FuzzyValues are defined but only between FuzzyValues that share the same FuzzyVariable This is reasonable because doing the intersection 268 CHAPTER 16 Writing the rules of hot temperature and low pressure makes no sense it s like adding 5 apples to 6 oranges FuzzyRules Let s continue the example and add two simple FuzzyRules which can be expressed in pseudo English as follows IF the temperature is warm THEN set the vent position to mostly open IF the temperature is hot THEN set the vent position to fully open The rules in FuzzyJess might appear like this defrule temp warm set vent mostly open temperature t amp fuzzy match t warm gt assert vent new FuzzyValue ventFVar mostlyOpen defrule temp hot set vent fully open temperature t amp fuzzy match t hot gt assert vent new FuzzyValue ventFVar fullyOpen The FuzzyJess function fuzzy match compares two FuzzyValues to see how well they match In this case the temperature fact has a FuzzyValue that is a fuzzified version of the current crisp temperature it is compared to the warm FuzzyValue If there is some degree of matching this level can be set by the user then the
127. the __data slot is normally hidden when the facts are dis played This is really just syntactic shorthand though ordered facts are unordered facts with a single multislot Jess understands that the name __data doesn t add any information and so should normally be hidden In practice the most common kind of ordered facts is the head only kind like initial fact They are often used for transient information that will soon be retracted and for one of a kind statements like shutdown now and found solution Ifyour fact will hold more than one piece of slot data unordered facts offer a lot more flexibility and lend structure to the information they hold so they are generally preferable in most situations In addition unordered facts lead to fewer coding errors The explicit slot labels serve as a sanity check during pro gramming and help prevent mistakes Finally unordered facts give better pattern matching performance as you ll learn in chapter 7 The show deftemplates function lists any implied deftemplates along with any explicitly created ones You can also use show deftemplates to see Jess s built in deftemplates Jess gt show deftemplates deftemplate MAIN __clear extends MAIN __fact Implied deftemplate MAIN __fact Parent template deftemplate MAIN __not_or_test_CE extends MAIN __fact Implied deftemplate MAIN initial fact extends MAIN __fact Implied deftemplate MAIN number extends MAIN __fa
128. the customer supplies her username A catalog screen where the customer chooses products to purchase A recommend screen where the customer selects additional recommended items to purchase A purchase screen where the customer checks out We re most interested in the catalog and recommend screens The Recommenda tions Agent won t be able to recommend anything until the customer has selected some items so you ll work on the catalog screen first Behind each of the screens is a servlet You ll develop the servlets one at a time and as we go along Pll explain the pieces of the Servlet API and J2EE as necessary 332 CHAPTER 19 Deploying web based applications 19 5 Starting the Catalog serviet 19 5 1 The Catalog servlet should perform the following steps 1 Get the login name previously stored by the Login servlet 2 Call get new order number to get a unique order number 3 Create a new user session with this login name and order number 4 Use the all products query to get a list of product facts 5 Display an HTML form to let the customer purchase products The last item should make you a little uncomfortable especially if you ve done web development in the past The code for HelloJess was to put it delicately ugly Java code to generate HTML is clumsy and verbose Worse if HTML is gener ated by Java then a Java programmer is required if it becomes necessary to make any changes to the HTML And changes w
129. the expression is of type RU LONG otherwise returns FALS Ea lowcase lt lexeme gt Converts uppercase characters in a string or symbol to lowercase Returns the argument as an all lowercase string matches lt lexeme gt Produces an ugly printout useful for debugging of the contents of the left and right memories of each two input node on the named rule or query s left hand side max lt number gt Returns the value of its largest numeric argument 404 APPENDIX A Jess functions member lt expression gt lt list gt Returns the position one based index of a value within a list otherwise returns FALSE min lt number gt Returns the value of its smallest numeric argument mod lt number gt lt number gt Returns the remainder of the result of dividing the first argument by the second modify lt fact id gt lt lexeme gt lt expression gt Modifies the given unordered fact in working memory Each of the lexeme expression lists is taken as the name of a slot in this fact and a new value to assign to the slot The numeric ID of the fact is preserved The jess Fact object is returned Modifying a shadow fact causes the appropriate object properties to be set as well See duplicate listp lt expression gt Returns TRUE for list values otherwise returns FALSE neq lt expression gt lt expression gt Returns TRUE if the first argument is not equal in type and value to any
130. these capabilities To accomplish this you ll add some new functions to the Jess language itself You ll define these functions in Java package them together and provide a simple mechanism for getting them loaded into Jess Once you ve done this the func tions will be indistinguishable from any of Jess s built in functions In this chapter I ll describe the general mechanism for adding new functions to the Jess lan guage and you ll apply it to the specific application The Userfunction interface In previous chapters you ve seen many examples of calling Java functions from the Jess language In chapter 13 you wrote an entire graphical interface this way The Jess language is wonderfully expressive when you re writing logical condi tions For many routine tasks though Jess language code is more verbose than the equivalent Java code would be There are even some things in particular manipulating multidimensional arrays that you can t do from the Jess language These tasks should be done in Java the old proverb about using the right tool for the right job comes to mind For these cases where writing Jess code alone isn t enough Jess includes the Userfunction interface If you implement Userfunction in a Java class Jess can invoke your class like any other Jess function In fact all of Jess s built in functions are themselves implementations of Userfunction The Userfunction interface is quite simple it contains o
131. this application you d want to try both multiple choice and numeric input and some recommendations that call halt and some that don t You can also run through some of the same test scenarios you used with the command line version of the application entering the data into the GUI instead Summary In part 4 of this book you ve built a fairly sophisticated application in about 230 lines of Jess code You collected knowledge in the form of flowcharts and then translated the flowcharts into rules The flowcharts are a compact and useful form of documentation for the resulting knowledge base The PC Repair Assistant uses backward chaining to automatically form hypoth eses about possible problems and ask the appropriate questions based just on the antecedents of the rules in the knowledge base This approach scales well to large systems unlike the more unwieldy approach used by the Tax Forms Advisor in part 3 Because you used backward chaining to let Jess choose the questions auto matically you needed only half as many rules compared to the tax advisor Once the application logic was written tested and debugged you wrapped it in a graphical user interface using only Jess code Adding a GUI required you to make some changes to the ask module used for asking questions but you didn t need to touch any of the rules The final incarnation of the ask module uses Jess s waitForActivations Java API function to communicate between Java
132. this discussion the two are equiv alent In both procedural and object oriented programming the programmer writes all the logic that controls the computer although it is partitioned differ ently in an object oriented program A purely declarative program in contrast describes what the computer should do but omits much of the instructions on how to do it Declarative programs 16 CHAPTER 2 What are rule based systems must be executed by some kind of runtime system that understands how to fill in the blanks and use the declarative information to solve problems Because declar ative programs include only the important details of a solution they can be easier to understand than procedural programs Furthermore because the control flow is chosen by the runtime system a declarative program can be more flexible in the face of fragmentary or poorly conditioned inputs when you removed some of the information from Mrs Rosencrantz s problem in chapter 1 the same pro gram unchanged was still able to find the possible solutions Declarative pro gramming is often the natural way to tackle problems involving control diagnosis prediction classification pattern recognition or situational aware ness in short many problems without clear algorithmic solutions Programming a cooking driving robot declaratively will be a breeze Although the driving program would be hard to write in a procedural style it is an ideal candidate to be wr
133. this second system will control the tem perature to within a small fraction of a degree Let s get started by developing a simple control algorithm for the HVAC system The control algorithm You can get an advanced degree in the engineering of control systems It s a vast and complex discipline and we can t hope to cover it all here Instead I ll state some commonsense principles to which your system should adhere and then we ll come up with a simple ad hoc control algorithm to satisfy these requirements Knowledge engineering with truth tables The best expert to interview for this project would be a control systems engineer Ideally he d already be familiar with heat pumps and have experience building control systems for them A certified Acme HVAC Systems engineer would of course be perfect The world is rarely perfect however and most likely you d talk to an engineer whose specialty wasn t HVAC control In that case you d combine knowledge from the expert with information gleaned from the Acme HVAC Systems Installation Guide the Acme User s Manual and any other documentation you could get your hands on Flowcharts are ideal for describing diagnostic systems but they would not be the best way to capture the collected knowledge here A better alternative is to use truth tables A truth table is a chart with rows and columns that correspond to some combination of inputs to the system The individual entries in the ta
134. to write a program so let s learn about function calls Later in this chapter we ll return to the study of lists in general Calling functions If you type an arbitrary list at the Jess gt prompt Jess will assume you re trying to call a function Jess gt 2 2 4 Here the function is being called Jess prints the result 4 on the next line The result is like the return value of a Java function The process of executing a func tion call to determine the result is called evaluating the function call Every Jess function has a result there are no void functions as there are in Java that don t have a return value Jess function calls use a prefix notation the head of the list is the name of the function being called and the other items in the list are the arguments of the function The arguments of a function call can be numbers symbols strings or other function calls that is function calls can be nested Jess gt 2 3 3 3 14 The two arguments to are 2 3 and 3 3 Jess evaluates the arguments in left to right order before the outer sum is computed Jess comes with a large number of built in functions that compute mathemati cal quantities control program flow manipulate strings give you access to Java APIs and perform other useful tasks Appendix A describes all the functions that are built in to Jess In section 4 4 you ll also learn how to define new functions by writing Jess language code
135. to all subsequent arguments For strings this means identical contents Uses the Java Object equals function so it can be redefined for external types Note that the integer 2 and the floating point number 2 0 are not eq but they are eq and eq lt expression gt lt expression gt Returns TRUE if the first argument is equivalent to all the others Uses numeric equality for numeric types unlike eq Note that the integer 2 and the floating point number 2 0 are not eq but they are eq and eval lt string gt Synonym for build evenp lt number gt Returns TRUE for even numbers otherwise returns FALSE Results with noninte gers are unpredictable exit Exits Jess and halts Java by calling System exit 0 exp lt number gt Raises the value eto the power of the argument explode lt string gt Creates a list value from a string Parses the string as if by a succession of read calls and then returns these individual values as the elements of a list external addressp lt expression gt Returns TRUE or FALSE depending on whether the given expression is an external address fact id lt integer gt If the argument is the fact id of an existing fact returns the actual jess Fact object otherwise throws an exception Jess functions 399 facts symbol Prints a list of all facts in working memory If an argument is given it should be the name of a module only the facts from that modul
136. to match any shopping cart that contains an odd number of items or a client that lives in a city whose name contains more than 10 letters Jess lets you specify these constraints and virtually any other constraint you can imagine using predicate functions For our purposes a predicate function is just a Boolean function a function that returns TRUE or FALSE You can use any predicate function as a constraint by preceding it with a colon If you want to use the value of a slot as an argument to the function and you almost always do you should bind that value to a variable first and then con nect that binding to the function using the amp connective Jess gt defrule small order shopping cart customer id id 106 7 2 5 CHAPTER 7 Writing rules in Jess contents c amp lt length c 5 checking out now id gt printout t Wouldn t you like to buy more crlf TRUE The lengths function returns the length of a list This rule delivers a special mes sage to any customers who go to the checkout with fewer than five items in their cart You can use the and or and not predicate functions to express complex logi cal conditions Although they are more verbose than the simple connective con straints they are more powerful because you can group them into arbitrary structures For example this rule fires if a customer is checking out with more than 50 items but his cart contains neither milk nor butter Jess g
137. to occur in response to a query you can include the max background rules declaration in that query s definition For example this query allows a maximum of five rules to fire while it is being executed Jess gt defquery find factorial declare max background rules 5 variables arg factorial arg Summary You can define rules to take action based on the contents of Jess s working mem ory and you can write queries to investigate it procedurally Both rules and que ries can use constraints conditions on the slot data of facts and conditional elements relationships between facts to express specify detailed requirements on working memory elements You can write both forward and backward chaining rules Roughly you can say that forward chaining rules discover the conclusions that can be drawn from an existing set of facts and backward chaining rules search for the premises from 132 CHAPTER 7 Writing rules in Jess which existing facts can be derived You can also write queries to probe working memory directly In real systems many rules are activated simultaneously Conflict resolution or choosing which rule to fire next is an important part of any rule based system Jess lets you influence conflict resolution in a number of ways by setting the conflict resolution strategy by using salience or by partitioning your rule base into modules Under the hood how Jess works In this chapter you ll
138. try to force rules to fire in a particular order If you find yourself using salience on most of your rules or if you are using more than two or three 7 6 7 6 1 Partitioning the rule base with defmodule 123 different salience values you probably need to reconsider whether you should be using a rule based approach to your problem If you want strict control over exe cution order then you re trying to implement a procedural program Either change your rules to be less sensitive to execution order or consider implement ing your algorithm as one or more deffunctions or as Java code Alternatively you might consider structuring your program using modules Partitioning the rule base with defmodule A typical rule based system can easily include hundreds of rules and a large one can contain many thousands Developing such a complex system can be a difficult task and preventing such a multitude of rules from interfering with one another can be hard too You might hope to mitigate the problem by partitioning a rule base into man ageable chunks The defmodule construct lets you divide rules and facts into dis tinct groups called modules Modules help you in two ways First they help you physically organize large numbers of rules into logical groups The commands for listing constructs rules facts and so on let you specify the name of a module and can then operate on one module at a time Second modules provide a con trol mechani
139. was being placed and then remove all the user s facts from the Rete object Deploying the application Deploying this web application is no more complex than deploying the single servlets you worked with before The files need to be installed like this TOMCAT lib jess jar webapps Order catalog jsp index html purchase jsp recommend jsp WEB INF web xml classes tekmart clp Catalog class Purchase class Recommend class 352 19 15 CHAPTER 19 Deploying web based applications NOTE The JSPs are all at the top level of the application The rule file tek mart clp is in the WEB INF classes directory so the class loader can find it One of the most confusing things about beginning servlet develop ment is remembering that the current directory of the servlet is likely to be the directory from which the servlet container was launched the serv let may not have read or write access to that directory As a result you can t expect to use the current directory to find files via relative paths The web xm1 file for the application is no more complex than the one for the first Hello servlet There is one servlet element and one servlet mapping tag for each servlet the servlet Catalog is mapped to the URL catalog and the map pings for the other two are similar The only new element is the context param element which is used to pass initialization parameters to the application lt context param gt lt p
140. way to incorporate this new feature using uzzy logic Controlling with fuzzy rules This section was written with Bob Orchard The first HVAC Controller system was based on traditional Boolean logic In Bool ean logic the state of a system is crisp meaning it has distinct well defined values For example in the system you ve developed the temperature of a floor is always in one of three distinct states too hot too cold or just right The decisions a Bool ean system can make are equally sharp a heat pump should be turned on or off and a vent should be open or closed Many modern control systems are based on fuzzy logic In fuzzy logic the same set of states can be used but the transitions between them are not sharp For example although temperatures above 72 degrees will still be considered too hot a fuzzy control system will consider 72 5 degrees to be slightly too hot and 16 6 1 Controlling with fuzzy rules 265 80 degrees to be very too hot A fuzzy control system can then make somewhat more subjective decisions Whereas the system you ve developed here only knows how to open or close the vent at each floor a fuzzy control system can continu ously vary the vent openings A fuzzy system will be able to control the tempera ture much more closely As an added benefit there will be less abrupt cycling of the heat pumps so they will have a longer useful life More generally fuzzy systems are subtler than Bo
141. we learn Where there is smoke there s fire and Murphy s Law Whatever can go wrong will go wrong are others that we learn throughout our lives Using this very general definition you might con clude that all the knowledge you have about the world can be encoded as rules Experience shows that this is often but not always the case In general any information you can think about in logical terms can be expressed as rules Rules are a lot like the 7f then statements of traditional programming lan guages You might write a gum chewing rule like this in an English like pseudocode IF I am in school AND I am chewing gum THEN spit out the gum END The if part of a rule written in this form is often called its left hand side often abbreviated LHS predicate or premises and the then part is the right hand side RHS actions or conclusions The domain of a rule is the set of all information the rule could possibly work with In this hypothetical case the domain of the chewing rule is a set of facts about the location and oral fixations of one particular person A rule based system is a system that uses rules to derive conclusions from premises Given the gum chewing rule and the premise that you are in school you as an advanced kind of rule based system yourself might conclude that it s time to spit out your gum In this book the systems we re talking about are a specific category One reviewer poin
142. with fuzzy rules 275 could also consider adjusting the guard temperatures or implementing a timer to make sure the heat pump does not cycle in less than some certain time The vent control rules used in this first fuzzy version are simple You have two basic situations to consider when the heat pump is cooling and when it is heating If it is off then it won t matter where the vents are set So when the heat pump is heating you define five rules that are related to the temperature and the terms defined for the temperature These rules should make some sense to you This is one of the useful features of fuzzy systems with a suitable set of concepts defined you can often express decisions in a simple and understandable way In pseudo English they look like this IF we are heating and the floor temperature is hot THEN set the vent to fullyClosed IF we are heating and the floor temperature is warm Then set the vent to slightlyOpen IF we are heating and the floor temperature is OK THEN set the vent to halfOpen IF we are heating and the floor temperature is cool THEN set the vent to mostlyOpen IF we are heating and the floor temperature is cold THEN set the vent to fullyOpen A similar set of five additional rules not shown here is defined for the situation where the heat pump is in a cooling state In FuzzyJess the code for one of the previous rules is written as follows defrule heating and temp cool HeatPump
143. write the code to do the work in Java and then call it from Jess Jess also has no special interface for working with multidimensional arrays so again Java code may be necessary You can either write ordinary Java functions and call them using the techniques from this chapter or you can use functions written in Java to extend the Jess language itself You ll learn how to do this in chapter 15 The JComboBox class has several constructors but in the previous example Jess knew which one you wanted to call based on the list you passed as an argument Let s see how Jess decides which method to invoke based on the function calls you write How Jess chooses among overloaded methods Jess is much less picky about data types than Java is In Java you can t store a float into a HashMap but you can store a Jess loat because Jess obligingly converts the number you provide into a java lang Double automatically and that can be stored just fine Most of the time these automatic conversions are helpful but occasionally they cause problems One problem area is when you need to call one of a set of overloaded Java methods A Java method name is said to be overloaded if multiple methods with the same name but different argument lists are available on the same object The many overloads of java io PrintWriter println are an extreme example All these methods appear in the PrintWriter class void printin void println boolean x void println ch
144. you could only extend the application by modifying the rules On the other hand the generic rules might be hard to understand and that would itself make the code more difficult to modify For the small set of forms this application will work with I think hard coding the tax information will lead 166 10 5 10 6 CHAPTER 10 Designing the application to a cleaner simpler application If you needed to work with 100 forms or 1 000 the other approach would be worth considering For this system though you won t need credit or deduction facts all the tax laws will be encoded directly in the rules Organizing the rules You ve defined five templates to serve as data structures for the application Now let s turn our attention from data to actual code The first order of business is to sketch out a rough structure for how the rules will be organized The Tax Forms Advisor needs to do four things 1 Initialize the application 2 Conduct an interview with the user to learn about her tax situation 3 Figure out what tax forms to recommend 4 Present the list of forms to the user removing any duplicate recommenda tions in the process These four steps map nicely onto four separate processing phases each with an independent set of rules You can put the rules for each phase into a separate defmodule as described in section 7 6 and take advantage of the support Jess offers for partitioning a problem into steps The four modules
145. you trigger the CLEANUP module to delete any previous partial order with this number Calling run lets the engine fire the CLEANUP rules Next the Recommend servlet needs to compose the order fact Several jess Value objects are needed for slot values so they re constructed first Next a Fact object is created and its slots are filled with setSlotValue Finally the com pleted fact is asserted with assertFact int orderNumber Integer parseInt orderNumberString Value orderNumberValue new Value orderNumber RU INTEGER Value customerIdValue new Value customerIdString RU ATOM Fact order new Fact order engine order setSlotValue order number orderNumberValue order setSlotValue customer id customerIdValue engine assertFact order Now that the order is in place the Recommend servlet builds the individual line item facts to represent the products being purchased for int i 0 i lt items length i Fact item new Fact line item engine item setSlotValue order number orderNumberValue item setSlotValue part number new Value items i RU ATOM item setSlotValue customer id customerIdValue engine assertFact item Note that it s OK to use the Value objects you created earlier because Value objects are immutable the data in a Value object can t be changed Getting the recommendations Once all the line items are asserted the Recommendations Agent can be invoked C
146. 03 Step Ctrl T Clear Hello World crif Halt Version watch activations watch compilations Figure 3 2 The JessWin developer s environment for Jess JessWin is a third party add on that provides a graphical interface to many functions that are useful to Jess programmers 3 2 Jess applications Jess has been used to develop a broad range of commercial software including m Expert systems that evaluate insurance claims and mortgage applications Agents that predict stock prices and buy and sell securities m Network intrusion detectors and security auditors m Design assistants that help mechanical engineers Smart network switches for telecommunications m Servers to execute business rules m Intelligent e commerce sites Games You ll develop some fairly large applications in this book the Tax Forms Advisor an intelligent information kiosk the PC Repair Assistant a graphical help desk application an HVAC Controller a soft real time control system and the Rec ommendations Agent a smart e commerce web site 3 2 1 Jess applications 37 You can program with Jess in two different but overlapping ways First you can use Jess as a rule engine A rule based program can have hundreds or even thousands of rules and Jess will continually apply them to your data Often the rules represent the heuristic knowledge of a human expert in some domain and the knowledge base represents the state of an evolving
147. 1 grocery list peas Now you add some advice to nth and then call it again Jess gt defadvice before nth Strip off the function name bind tail rest argv Convert zero based to one based index bind index 1 nth 1 tail Put the argument list back together bind Pargv create nth index rest tail TRUE Jess gt nth 1 grocery list beans The result is different this time the index is interpreted as one based and the answer is beans instead of peas The nths function will continue to behave this way until Jess is restarted the clear function discussed in chapter 6 is called or the advice is removed with undefadvice Jess gt undefadvice nth TRUE Jess gt nth 1 grocery list peas You may be surprised that you were able to call nths in the previous advice code It looks as though you re redefining nth in terms of itself a recipe for an infinite loop To prevent this kind of problem Jess doesn t apply advice to any function calls invoked in an advice block That s why your call to nths inside the advice block uses a one based index In the nth example the advice code executes before the built in nths func tion is invoked If the first argument to defadvice were after instead of before the advice would execute after nths In this case the special variable retval is also defined it holds the result returned by the adviced function The advice code can modify this varia
148. 2 lowcase 403 Macintosh 191 MAIN 123 166 module 78 main module 123 main thread 209 Makefile 328 330 428 manual testing of GUIs 216 manufacturing resource plan ning 288 marketing 293 matches 146 403 mathematical functions 164 max 403 member 52 60 106 404 membership in a fuzzy set 266 INDEX 439 memory usage during pattern matching 297 message driven beans 376 metadata 366 Microsoft 324 Microsoft Word 358 min 404 missing information 7 mobile agents xxi mod 404 modal dialog 214 model view controller paradigm See MVC modify 85 94 107 238 404 modifying facts 85 modular applications 184 modularity 202 modules 26 78 123 166 173 defining 124 focus 126 name lookup 125 name resolution 124 monographs 156 motherboard 197 moving expenses 176 MRP 222 multifields 57 103 multiple catch blocks 72 multislots 84 89 195 294 313 creating from Java 313 multithreading 264 MULTIVARIABLE 422 Murphy s Law 17 mutators 66 MVC 332 MYCIN 18 188 N namespaces 124 native Java keyword 225 native methods 224 negation count 142 neq 60 404 Netscape 289 440 INDEX new 62 307 404 newline 44 newsletters 156 Newton s Law of Cooling 227 nil 43 83 119 164 245 node index hash value 141 node sharing 139 143 NodelMTEQ 145 NodelNONE 145 Node1RTL 145 NodelTECT 145 NodeNot2 145 NodeTerm 145 NodeTest 145 nonlinearity 223 non public classes 134 not 404 nested 113 predicate function 10
149. 4 lt Fact 3 gt Jess gt focus report MAIN Jess gt run Please take a copy of form f1 Explanation ef1 Please take a copy of form f2 Explanation ef2 Please take a copy of form f3 Explanation ef3 Please take a copy of form f4 Explanation ef4 4 Jess gt Note that the forms are listed in alphabetical order and each one is displayed with its explanation The application is now basically done you just need to han dle a few remaining details Finishing touches Now that all the rules are written you need a way to start your program Because the module focus must be shifted several times during each run of the applica tion it is helpful to define a deffunction to perform the proper sequence of operations deffunction run system reset focus startup interview recommend report run The single call to focus sets up the focus stack so the four modules become active in the right order The program can then be executed with a single call to run system This application is intended to run continuously in a kiosk at the post office so you call this function in a loop 11 6 Testing the full application 183 while TRUE run system This loop restarts the application each time it terminates Testing the full application Now that the whole application is written dust off the sample scenarios you devel oped for testing the interview and recommend modules and run them again
150. 6 NRC 265 nre fuzzy 269 nth 50 58 404 null 43 245 null fact 142 NumberFormatException 71 numberp 168 405 o OBJECT slot name 90 object nursery size 39 oddp 405 one input nodes 137 open 405 OPS5 32 136 or 405 conditional element 109 implementation 110 predicate function 106 Orchard Bob xxiii 219 265 order configuration 288 ordering patterns 297 oscillations 262 OutOfMemoryError 143 overshoot 281 P package names 204 packages 225 247 253 page testing 343 partial matches 297 pattern bindings 100 107 negated patterns 112 with exists 113 pattern matching 19 22 134 efficient 136 inefficient 135 performance of 297 pattern nodes 138 patterns 98 payroll 358 PC Repair Assistant xxvii 185 peer review 156 pencil 152 189 phases 26 physics 223 pi 405 pickling 235 plain lists 67 policy file 382 pop focus 127 405 possibility distribution 266 POST 338 ppdeffacts 126 405 ppdeffunction 405 ppdefglobal 405 ppdefquery 405 ppdefrule 97 126 130 405 return value 98 ppdeftemplate 86 405 predicate constraints 105 predicate functions 105 113 predicates 17 102 premises 17 75 price list 64 principles 151 printout 35 47 167 319 320 329 406 PrintWriter 68 326 priority 275 procedural language 96 procedural programming 14 15 progn 53 406 property change events 264 PropertyChangeEvent 230 234 315 PropertyChangeListener 230 PropertyChangeSupport
151. 7 2 Constraining slot data 100 Literal constraints 101 Variables as constraints 102 Connective constraints 104 Constraining matches with predicate functions 105 Return value constraints 106 Pattern bindings 107 7 3 Qualifying patterns with conditional elements 108 The and conditional element 108 The or conditional element 109 The not conditional element 111 The test conditional element 113 The logical conditional element 114 7 4 Backward chaining rules 116 7 5 Managing the agenda 120 Conflict resolution 121 Changing rule priority with salience 121 7 6 Partitioning the rule base with defmodule 123 Defining constructs in modules 123 Modules scope and name resolution 124 Module focus and execution control 126 7 7 Searching working memory with defquery 128 The variable declaration 130 Query trigger facts 130 The count query results function 131 Backward chaining and queries 131 7 8 Summary 131 Under the hood how Jess works 133 8 1 Review ofthe problem 134 8 2 An inefficient solution 135 8 3 The Rete algorithm 136 How Rete works 136 Handling retract 139 8 4 Easy optimizations for Rete 139 8 5 Performance of the Rete algorithm 141 Node index hash value 141 CONTENTS xiii 8 6 More complexity and initial fact 142 Implementing the not conditional element 142 Implementing the test conditional element 143 Implementing backward chaining 143 8 7 Exploring the Rete network in Jess 143 The watch com
152. 9 JavaServer Pages Another way to deploy server side Java applications is via JavaServer Pages JSPs A JSP is an HTML page with embedded Java code In a way a JSP is an inside out version of a servlet A servlet is Java code that often includes statements that print HTML JSPs are compiled by a special program on the web server generally into servlets and then executed in response to browser requests Whereas servlets are ideal when a server side component needs to do a nontrivial amount of work and produce only a small amount of HTML JSPs are perfect when a large HTML page needs to embed a small amount of computed information Servlets and JSPs are often used together with the JSPs providing the interface and the servlets provid ing the logic You ll use JSPs to provide the user interface for the Recommenda tions Agent in chapter 19 Perhaps the best thing about JSPs is that they can be written in a regular HTML editor Servlets must be written by experienced Java programmers but JSPs can be written by web designers perhaps with the use of some boilerplate provided by a Java guru It s difficult to maintain a web page created by a servlet using a series of print1n calls especially for a nonprogrammer but maintaining a JSP based site is not very different than maintaining a site built from static HTML pages Web services Servlets and JSPs are designed to deliver HTML interfaces to human users Web services on the other hand are d
153. 95 May 2003 The comprehensive resource for the practicing EJB developer who wants quick clean solutions to frequent problems It is a systematic collec Benjamin G Sullins Mark B Whipple tion of EJB recipes Each recipe describes a practical problem and its background then shows the code that solves it No EJB developer should be without this book BH ETTIEG For easy ordering call 1 800 247 6553 or fax 1 419 281 6883 For online browsing secure ordering special promotions and fast delivery visit www manning com EXPERT SYSTEMS JAVA JESS IN ACTION Ernest Friedman Hill Rule Based Systems in Java and facts instead of the usual linear set of instructions That s the idea behind rule based programming A rule engine automati cally decides how to apply the rules to your facts and hands you the result This approach is ideal for expressing business rules and is increasingly used in enterprise computing magine a different way to program in which you specify rules Jess is a popular rule engine written in Java Its supported by Sandia Labs and has an active online community If you have a problem that can be solved with rules Jess in Action will show you how If you are not sure read chapter 2 Written by the creator of Jess this book is an accessible and practical guide to rule based system development in Java Jess in Action first introduces rule programming concepts and teaches you the Jes
154. CLEAR 139 RU EXTERNAL_ADDRESS 62 249 309 RU FACT 78 RU FLOAT 43 RU FUNCALL 243 RU INTEGER 43 63 195 RU LIST 313 RU LONG 43 RU MULTIVARIABLE 242 RU REMOVE 139 RU STRING 44 RU UPDATE 139 RU VARIABLE 242 rubber ducks 295 rule actions in XML 365 rule base 19 21 rule compiler 21 rule development and deploy ment environment 18 rule engines 16 18 21 37 standards 26 rule languages common concepts 361 Jess 32 standard 27 vendor specific 27 rule representations 361 rule based systems 17 architecture 19 developing 23 RuleML 363 364 rules 17 407 backward chaining 96 116 compiling 99 example 137 editors 362 366 372 firing 98 forward chaining 96 interoperability 361 Jess function 130 rewriting 110 server 379 simplest possible 96 storage 362 writing 25 XML representation 365 run 98 100 170 210 311 407 in multithreaded applications 346 Runnable 273 runQuery 340 run query 129 130 301 407 runtest ref 429 INDEX 441 runtest sh 429 run until halt 201 232 408 S salience 121 275 281 performance impact 122 sample code 263 Sandia National Laboratories xxi 32 save facts 408 scalability 193 358 script UNIX command 177 scripting 62 seamless 359 search engines 157 searchability 361 security 352 selection sort 181 Serializable 235 serialization 235 247 and RMI 380 servlet container 290 326 servlet jar 327 ServletContext 338 340 servlets 290 291 325 377 cookies 339 deployin
155. ED 76 3857453257847 OFF CLOSED 763944595207749 pump state vent state and temperature OFF CLOSED 76 40320240178788 for each floor The text field on the OFF CLOSED 76 41197408445163 bottom shows the current outdoor temperature The simulator is now a reasonable stand in for a real HVAC system Next you need a way to connect the simulator or in its place the HVAC system itself to Jess You ll do this by wrapping the simulator s interface in a set of JavaBeans Writing the JavaBeans Jess s working memory can hold not only plain facts but also shadow facts which are placeholders for Java objects outside of Jess see section 6 5 for the details More specifically they are placeholders for JavaBeans From Jess s perspective a JavaBean is just an instance of a Java class with one or more properties specially named methods that let you read and or write to a named characteristic of the object A very simple JavaBean to represent a temperature sensor for the HVAC system could look like this 230 CHAPTER 14 The reality connection package control public class Thermometer private Hardware m_hardware private int m_floor public Thermometer Hardware hw int floor m_hardware hw m_floor floor public double getReading return m_hardware getTemperature m_floor This Bean has one read only property named reading If you used this as is Jess could create shadow facts for Thermometer
156. FactoryImpl extends UnicastRemoteObject implements JessFactory private JessFactoryImpl throws RemoteException public Jess create throws RemoteException return new JessImpl1 Remote object implementations often extend the UnicastRemoteObject class but this is not required By extending UnicastRemoteObject a remote object implementation class gains the ability to automatically export itself or connect to a network Because UnicastRemoteObject s constructor throws RemoteException JessFactoryImp1 s constructor must throw it as well The implementation of the single remote method is about as simple as you can get Improvements might include a pooling mechanism create could check a collection of previously cre ated Rete objects before making a new one Here s an implementation of the Jess remote interface it s only slightly more complex than JessFactory import java rmi import java rmi server import jess public class JessImpl extends UnicastRemoteObject implements Jess private Rete m_rete new Rete JessImpl throws RemoteException public Value executeCommand String command throws JessException return m_rete executeCommand command public void store String name Object object m_rete store name object public Value fetch String name return m_rete fetch name Each instance of JessImp1 contains a single instance of jess Rete so JessImp1 obj
157. For Java objects the toString method is called str compare lt lexeme gt lt lexeme gt Lexicographically compares two strings Returns 0 if the strings are identical a negative integer if the first is lexicographically less than the second or a positive integer if the first is lexicographically greater Jess functions 411 str index lt lexeme gt lt lexeme gt Returns the position of the first argument within the second argument This is the one based index at which the first string first appears in the second otherwise returns FALSE stringp lt expression gt Returns TRUE for strings otherwise returns FALSE str length lt lexeme gt Returns the length of the argument in characters subseq lt list gt lt integer gt lt integer gt Extracts the specified range from a list consisting of the elements between the two one based indices of the given list inclusive and returns this range as a new list subsetp lt list gt lt list gt Returns TRUE if the first argument is a subset of the second that is all the ele ments of the first list appear in the second list otherwise returns FALSE sub string lt integer gt lt integer gt lt string gt Retrieves a subportion from a string Returns the string consisting of the charac ters between the two one based indices of the given string inclusive symbolp lt expression gt Returns TRUE for symbols otherwise returns FALSE sym cat lt expres
158. HOT temp RATE break case COOLING temp COLD temp RATE break case OFF temp i 1 0 005 break Outdoor leakage temp m_outdoor temp RATE 2 m_temperature i temp try Thread sleep 1000 catch InterruptedException ie return 14 2 4 14 3 Writing the JavaBeans 229 private int state int floor if getVentState floor 1 return getHeatPumpState floor 3 1 else return OFF Adding a graphical interface Now the simulator runs but currently there s no way to see what it s doing A graph ical display that shows the full state of the building at a glance would be very help ful A simple GUI could display a row of labels for each floor one label each for heat pump state vent state and temperature If you add a text field that lets you set the outdoor temperature the simulator will be ready to test the HVAC Controller system Figure 14 2 is a screen shot of this simple interface I won t show the code here but it s included in the sample code for this chapter The SimulatorGUI class has a constructor that accepts a Simulator as a constructor argument and then uses a thread to poll the Simulator to determine its state over time OFF CLOSED 76 34 26006327599 OFF CLOSED 76 35117311279437 Figure 14 2 OFF CLOSED 76 35977370895189 OFF CLOSED 76 36840253412377 A GUI for the HVAC simulator The three OFF CLOSED 76 3770597017423 columns of labels represent the heat OFF CLOS
159. HVAC system in real time The sample rule in section 14 3 1 suggests how these JavaBeans will be used In the next chapter you will write additional Java functions to help control the HVAC systems You will then use these Java functions to extend the Jess language by adding new commands While we re at it we ll look at Jess s extension facilities in general and the jess Userfunction interface in particular and you ll learn how to write a range of different Jess extensions 236 CHAPTER 14 The reality connection Armed with all the Java code developed in this and the next chapter writing the rules to control the HVAC system will be straightforward That will be the task of chapter 16 Extending the fess language In this chapter you ll Be introduced to the Userfunction interface Learn to extend the Jess language a Write Jess functions to interface with HVAC hardware 237 238 15 1 CHAPTER 15 Extending the Jess language The HVAC Controller system you re writing needs to be able to switch the various heat pumps and vents on and off in response to changing temperatures Although it would be possible to do this just by using the modify function on the shadow facts representing the devices that approach won t always be convenient Further more the rules need access to other functions in the Hardware interface that aren t used by the JavaBeans It would be best to have a simple and consistent interface to all
160. ISBN 1 930110 89 8 Printed in the United States of America 12345678910 VHG 06 05 04 03 To my family brief contents PART 1 INTRODUCING RULE BASED SYSTEMS 1 1 Rules to the rescue 3 2 What are rule based systems 13 PART 2 JESS A RULE BASED PROGRAMMING ENVIRONMENT 29 Introducing Jess 31 Getting started with the Jess language 41 Scripting Java with Jess 61 3 4 5 6 Representing facts in Jess 74 7 Writing rules in Jess 95 8 Under the hood how Jess works 133 vii Vili BRIEF CONTENTS PART 3 CREATING YOUR FIRST RULE BASED PART 4 PART 5 PART 6 PART 7 APPLICATION THE TAX FORMS ADVISOR 147 9 Collecting the knowledge 149 10 Designing the application 161 11 Writing the application 172 WRITING A DIAGNOSTIC APPLICATION THE PC REPAIR ASSISTANT o on 185 12 Writing the PC Repair Assistant 187 13 Adding a graphical interface 203 REASONING ABOUT REALITY THE HVAC CONTROLLER 0 0cccceeeeeees 219 14 The reality connection 221 15 Extending the Jess language 237 16 Writing the rules 255 TEKMART COM RULE BASED APPLICATIONS FOR THE WEB sicvcsiccsatocionsovcssn acaconcsstuctiviies 285 17 Jessonthe Web 287 18 Embedding Jess in Java applications 306 19 Deploying web based applications 323 ENTERPRISE SYSTEMS 0 cccccsseccceeeeeeeeeees 355 20 Jess XML and the enterprise 357 21 Jess i
161. If there are any honest men in the world the rule will fire once and only once The exists CE is implemented as two nested not CEs that is exists A is exactly the same as not not A Therefore you can t bind any variables in an exists CE for use later in the rule and you also can t use pattern bindings with exists The test conditional element A pattern with test as the head is special the body consists not of a pattern to match against the working memory but of a Boolean function The result deter mines whether the pattern matches A test pattern fails if and only if the function evaluates to the symbol FALSE if it evaluates to TRUE the pattern succeeds For example suppose you wanted to find people whose age is less than 30 years old Jess gt deftemplate person slot age TRUE Jess gt defrule find trustworthy people 1 person age x test lt x 30 gt printout t x is under 30 crlf TRUE 114 7 3 5 CHAPTER 7 Writing rules in Jess A test pattern like a not cannot contain any variables that are not bound before that pattern and it can t have a pattern binding Because a test CE like a not CE doesn t match an actual fact its implementa tion is similar to the way not is implemented A test CE is evaluated every time the preceding pattern on the rule s LHS is evaluated just like a not Therefore the following rule is equivalent to the previous one Jess gt defrule find trustw
162. Impl_Skel class and the stub and skeleton classes for JessFactoryImpl The RMI rule server is now ready to launch First run the rmiregistry pro gram and then execute the JessFactoryImp1 class with your JVM You should see the Jess server running message if everything worked OK A sample client You re done writing the rules server and it s ready to be used from an EJB acting as a client Although you won t be writing any EJBs here of course you want to see what the client code would look like Here s a simple client application that builds a Jess command from its arguments and executes the result in a remote Jess object import java rmi public class JessClient 21 2 6 21 3 JSR 94 the javax rules API 383 public static void main String argv throws Exception JessFactory factory JessFactory Naming lookup Jess Jess jess factory create StringBuffer sb new StringBuffer for int i 0 i lt argv length i sb append argv il sb append System out println jess executeCommand sb toString You need the two remote interface classes to compile this client and to run it you need the two stub class files as well Here s a sample run java JessClient 2 3 uo Final polishing We ve ignored many of RMI s subtle points in this section For example RMI includes a mechanism for loading stub class files and other classes remotely using a web server as a c
163. Java method this way There are a few special cases Static methods methods returning or accepting arrays and overloaded methods all require special care We ll discuss these cases in the next few sections First let s look at a shortcut that will improve the readability of the last example Nesting function calls and a shortcut The symbol ca11 in the following example is a little distracting call prices put bread 0 99 It is actually no more verbose than the equivalent Java code map put bread new Double 0 99 In fact the Jess code is a little shorter But still call seems like extra baggage The good news is that Jess lets you omit it prices put bread 0 99 When the first element of a function call is a Java object Jess assumes you meant to include the symbol ca11 and invokes a function on the object This works even if the first element of the function call is another function call bind prices new HashMap put bread 0 99 This single line of code creates a HashMap binds it to a variable and adds a name value pair Be careful with nesting function calls this way however combining logi cally separate operations into one line of code can make your programs hard to understand Used wisely though such compact code can be readable and efficient 66 CHAPTER 5 Scripting Java with Jess For most method calls the call is optional However you can t leave it out when you re calling a static method 5
164. Jess for educational purposes so you can try out and experiment with the exam ples in this book In this chapter I ll assume you re using that version of Jess Jess is also available from the Jess web site http herzberg ca sandia gov jess There you can immediately download a trial version or you can obtain a Jess license and then download a full version with source code Jess licenses are avail able free of charge for academic use You can also purchase a commercial license See the web site for details Installing Jess To run Jess you need to have the Java 2 Platform installed If you don t already have it you can download a free implementation directly from Sun Microsystems at http java sun com Versions are available for Windows Linux and Solaris Apple has its own version which is included in Macintosh OS X If Java is prop erly installed you should be able to execute the command java version in a command window and see version information about your installed Java soft ware If the version number is 1 2 0 or greater you re ready to go The Jess distribution is a Zip file and you can open it with many popular archiving tools WinZip http www winzip com is one useful program for work ing with Zip files on Microsoft operating systems The Zip file contains a single item a directory named Jessxx where XX is a number currently 61 for Jess 6 1 Use your archiving tool to unpack that directory to a convenient locat
165. Main taxes clp The clp extension is traditional but not required Using a specific extension con sistently is helpful because you may be able to train your programmer s editor to recognize Jess code by the filename Organizing the data As you know Jess rules work by pattern matching on facts Therefore before you can write any rules you need to have some idea what the facts will look like Of course in one of those classic chicken and egg problems you don t know what the facts should look like until you see the rules How do you get started Generally the knowledge engineering effort suggests some possible fact cate gories If you record the knowledge as proposed rules or rule like pseudocode perhaps using the index card method described in chapter 9 the possible fact types will be explicitly laid out Otherwise you ll have to read through the col lected knowledge to get a feel for the kinds of facts that are required The whole process is subjective and there is no right answer With practice you ll get a feel ing for what will work and what will not Looking through chapter 9 s collected knowledge for the Tax Forms Advisor you can see some possible candidates for deftemplate types form A specific tax form user The operator of the system 10 2 10 2 1 Filling in details 163 deduction A way of reducing your taxable income credit A way of reducing your tax burden dependent A per
166. ND customer type is wholesaler THEN deduct 10 percent from order END If a company s business rules are implicit not written as rules per se but embed ded in procedural logic and scattered throughout corporate computer applica tions then a change in a single policy might require significant programmer effort to implement Furthermore if business rules are to be embedded directly into application software it becomes difficult to use commercial off the shelf COTS products increasing the company s development costs The corporation will be forced to make a choice between containing development costs and mak ing policy adjustments in response to changing circumstances The solution to this dilemma is to remove the business rules from the individ ual applications make them explicit and embed them in a centralized rule engine for execution Any business policy can then theoretically be changed at a single point The rule engine is often embedded in a network based server so that it can be accessible across an enterprise This enterprise level use of a rule engine is probably the fastest growing and most visible market for rule based systems programming today Some application servers like BEA s WebLogic include integrated rule engines Other vendors like ILOG sell rule engines meant to be used with third party servers There are literally 12 13 CHAPTER 1 Rules to the rescue dozens of rule engines to choose from t
167. Nothing Do Nothing Do Nothing Start Cooling Heating Do Nothing Do Nothing Do Nothing Turn Off Turn Off Cooling Turn Off Turn Off Do Nothing Do Nothing Do Nothing Saving energy There are two ways to shut off the air to a given floor You can close the vent for that floor or shut off the heat pump that services it If any floor serviced by a heat pump needs to be heated or cooled then of course that heat pump needs to be turned on If a heat pump is on and none of its floors needs service then you must be sure to shut off the heat pump rather than simply closing all the vents Setting up It s time to begin writing the Jess code for the HVAC Controller Before you do anything else load the Userfunctions you wrote in chapter 15 load package control HardwareFunctions Next you need to create a Simulator object and make it available to Jess You can store a reference to it in a defglobal using code like this defglobal hw init simulator 9 set reset globals FALSE The graphical monitor for the simulator will appear as soon as the first line is exe cuted The call to set reset globals tells Jess not to reinitialize hw when reset is called If you didn t call set reset globals you could end up with mul tiple simulator GUIs on the screen and the connections between the JavaBeans and the simulator would be broken Next you need to use defclass to tell Jess about your JavaBean classes and then c
168. R 11 Writing the application Please take a copy of form 2441 Explanation Child care expenses Please take a copy of form 8829 Explanation Home office expenses Now would be a good time to take the system to the tax expert to let her play with it She very well may have some important changes or additions you need to inte grate it is the nature of software development that no set of requirements is ever complete You should also show this system to the customer to get some feedback Summary Over the last three chapters you ve gone from concept to research to design and finally to execution of a system for recommending tax forms Although the pro gram you ve developed is somewhat simplistic the process you followed was repre sentative of rule based system development And although the user interface of the application you developed here is unsuitable for real use the underlying logic is perfectly respectable and includes features like input validation and a simple explanation facility The routines you developed for asking questions and validat ing input will serve as a foundation on which you ll build more sophisticated tools later in this book In this chapter you saw how a good design and sturdy infrastructure help sim plify programming a system You learned to write a modular application using processing phases implemented using defmodules and you got some practice writing simple pattern matching expressions You
169. RFuzzySet 0 10 0 0 new RightLinearFunction m_thermometerRateOfChangeFV addTerm zero new TriangleFuzzySet 0 10 0 0 10 m_thermometerRateOfChangeFV addTerm increasing new LFuzzySet 0 0 0 10 new LeftLinearFunction catch XValuesOutOfOrderException outOfOrderEx NOTHING catch XValueOutsideUODException outsideOUD NOTHING public FuzzyValue getFuzzyReadingRateOfChange return m_fuzzyReadingRateOfChange null 120 Set up FuzzyVariable 10 27 JavaBeans property for temperature change rate 278 CHAPTER 16 Writing the rules public static void SetThermometerFVTerms double setPoint try m_thermometerFV addTerm cold new RFuzzySet setPoint 1 5 setPoint 1 new RightLinearFunction m_thermometerFV addTerm cool new TriangleFuzzySet setPoint 1 5 setPoint 1 setPoint m_thermometerFV addTerm OK new TriangleFuzzySet setPoint 1 setPoint setPoint 1 m_thermometerFV addTerm warm new TriangleFuzzySet setPoint setPoint 1 setPoint 1 5 m_thermometerFV addTerm hot new LFuzzySet setPoint 1 setPoint 1 5 new LeftLinearFunction catch XValuesOutOfOrderException outOfOrderEx NOTHING catch XValueOutsideUODException outsideOUD NOTHING public void run Method to move boolean firstLoopDone false temperature set point while true double reading getReading double rateOfChange
170. Rule Based Systems in Java IN ACTIO Ernest triedman Hill WE vannine Jess in Action Jess in Action Rule Based Systems in Java ERNEST FRIEDMAN HILL LL MANNING Greenwich 74 w long For online information and ordering of this and other Manning books go to www manning com The publisher offers discounts on this book when ordered in quantity For more information please contact Special Sales Department Manning Publications Co 209 Bruce Park Avenue Fax 203 661 9018 Greenwich CT 06830 email orders manning com 2003 by Manning Publications Co All rights reserved No part of this publication may be reproduced stored in a retrieval system or transmitted in any form or by means electronic mechanical photocopying or otherwise without prior written permission of the publisher Many of the designations used by manufacturers and sellers to distinguish their products are claimed as trademarks Where those designations appear in the book and Manning Publications was aware of a trademark claim the designations have been printed in initial caps or all caps Recognizing the importance of preserving what has been written it is Manning s policy to have the books they publish printed on acid free paper and we exert our best efforts to that end Manning Publications Co Copyeditor Tiffany Taylor Mi 209 Bruce Park Avenue Typesetter Syd Brown Greenwich CT 06830 Cover designer Leslie Haimes
171. T FUNCALL INTEGER LIST LONG MULTIVARIABLE STRING and VARIABLE B 9 1 B 10 B 11 jess Value 423 public static String gensym String prefix Returns a unique symbol starting with the given prefix This method is useful if your Java code is automatically generating Jess rules variables or other entities and needs to come up with a unique name jess Token The Token class represents a group of facts as they travel together through Jess s Rete network The iterator returned by Rete runQuery or the run query func tion in Jess returns a series of Token objects otherwise you shouldn t need to deal with this class public Fact getFact int index Returns a single Fact from the token The fact that matches the first pattern in a rule has index 0 public int size Returns the number of facts in a Token public Fact topFact Equivalent to token getFact token size 1 jess Value This is an important class because every piece of information that travels through Jess a number a string a Java object does so wrapped in an instance of jess Value or one of its subclasses Nevertheless I won t say much about it here because it is covered well in chapter 15 Note that the type constants are defined in the class RU public Value boolean b Constructs a Boolean value object one of the symbols TRUE or FALSE wa public Value double d int type Constructs a numeric value The type can b
172. The javax rules API also doesn t provide an escape mechanism by which vendors can expose nonstandard interfaces although many of its methods accept prop erty lists as a customization mechanism Jess is more than just a rule engine so the javax rules interface limits what you can do with Jess Implementations of JSR 94 intended for use with J2EE containers deal with the programming restrictions imposed by that environment For example some JSR 94 implementations are built on top of an RMI service like the one you built in the last section Other JSR 94 implementations are targeted toward high perfor mance in J2SE applications and some can switch between the two At the time of this writing the JSR 94 specification is being readied for public release and the only implementation is the reference implementation which incidentally is a driver for Jess Some of what I l present here will probably change but much of this material will still be valid when the javax rules API becomes official Working with javax rules The javax rules API divides interaction with rule engines into administrative and runtime interactions Administrative tasks include instantiating the rule engine and loading rules and runtime tasks include manipulating working memory and executing rules If you use javax rules from a J2SE program you ll probably need to perform all of these tasks from your code On the other hand in the J2EE environment the administrative t
173. Two kinds of experts might 17 4 Designing data structures 293 have knowledge you could add to such a system technical folks and marketing folks The technical knowledge will let the system tell the customer how to assem ble a working system from individual components The marketing knowledge can be used to try to convince the customer to buy more stuff For the current system you ll base technical recommendations on the concept of categories Every product will belong to one category to use it products from other categories may be required or desirable For instance the product Univac 2000 belongs to the category computer to use it you might need a video monitor category monitor because the monitor is not included A monitor is therefore a requirement for computer customers If the user of your system buys a Univac 2000 the system should recommend that she buy a monitor too Other accessories might also be nice speakers and software for example These should also be rec ommended and for the purposes of this system you ll also call them requirements The marketing derived rules that you ll include will be designed to help sell videotapes and video discs DVDs Assume that the marketing folks have identi fied four opportunities in this area Ifa customer is buying a VCR or DVD player recommend the appropriate media Ifa customer has bought a VCR or DVD player on a previous visit recom mend more media for it
174. You can often tune the performance of a Jess application by tuning the behavior of the garbage collector The next five chapters will teach you the Jess language in detail First you ll learn to use Jess as a pure programming language and then you ll see how to write rules Along the way especially in chapter 8 we ll peer under the hood and see how Jess works Let s get started Getting started with the Jess language In this chapter you ll m Learn the basic syntax of Jess Work with basic Jess control structures m Find out how to define functions in Jess 41 42 4 1 4 1 1 4 1 2 CHAPTER 4 Getting started with the Jess language This chapter is an introduction to the Jess rule language It will not teach you how to write rules that will have to wait for chapter 7 but it will explain how to write code in the Jess rule language and you ll need this ability to write the right hand sides of rules The Jess rule language is also a general purpose programming lan guage and you can use it even without writing any rules The next few chapters contain syntax diagrams to concisely describe the syntax of some of the more complicated expressions In these diagrams m Text in lt angle brackets gt is a description of some kind of data that you must supply m Things ending with can appear one or more times m Things ending with can appear zero or more times m Things in square brackets are optional Square brac
175. _data new Value new ValueVector add new Value bread RU ATOM add new Value milk RU ATOM add new Value jam RU ATOM RU LIST engine assertFact f i This code uses an interesting shortcut The add method in ValueVector returns the ValueVector you call it on This means you can chain calls to this method as we ve done here Some people like this technique and some people hate it I think it s a fine thing to do when you re building up a list of items Removing facts The retract method in the Rete class lets you remove facts from working memory Fact f engine retract f retract takes a Fact object as an argument If you asserted the fact from Java then it s easy to use it as an argument to retract If you didn t then you need to get hold of a Fact object somehow There are two approaches Either you can use the fact number of the Fact as an argument to the findFactById method in Rete or you can construct a Fact just like the one you want to retract and use that as the argument to retract it turns out that retract doesn t need you to pass ina Fact object that s really in working memory it can instead be a Fact that s identi cal to one in working memory The working memory is stored as a hash table with the Fact objects as the keys That means finding out whether a Fact is in working memory already is a fast operation but finding the Fact with a particular numeric identifier is slow b
176. a 5 Nothing happens until the user enters some text and presses Enter or clicks the OK button 6 The event handler read input running on the event thread asserts an ask user input fact containing the text entered by the user as a symbol 7 The ask user input fact together with the MAIN question fact and the MAIN ask fact activate the rule collect user input The method waitForActivations finally returns due to this activation and the right hand side of ask quest ion by id completes Better input components 213 s Back on the main thread collect user input asserts a MAIN answer fact retracts the MAIN ask and ask user input facts and returns from the ask module 13 5 Better input components As it stands the user of your program has to answer questions by typing a reply into a JTextField Of course most GUIs don t work that way For example users aren t accustomed to typing x86 or Macintosh into a GUI they re used to select ing choices from a combo box This application should work the same way Recall that the question fact template has a slot named type indicating the kind of answer expected for each question Your interface should display different user input panels depending on the value of this type field What panels will you need Looking back at the questions defined in the previous chapter you ll need only two Multiple choice A panel with a pop up menu a combo box containing a list of c
177. a tion An algorithm that could somehow remember previous pattern matching results between cycles only updating matches for facts that actually changed could do far less work and get the same results Jess uses a very efficient version of this idea known as the Rete algorithm Charles Forgy s classic paper describing the Rete algorithm became the basis for several generations of fast rule based system shells OPS5 its descendant ART CLIPS Jess and others Each system has enhanced and refined the algorithm to improve performance or flexibility This chapter describes the algorithm as imple mented in Jess Briefly the Rete algorithm eliminates the inefficiency in the simple pattern matcher by remembering past test results across iterations of the rule loop Only new or deleted working memory elements are tested against the rules at each step Furthermore Rete organizes the pattern matcher so that these few facts are only tested against the subset of the rules that may actually match How Rete works Rete is Latin for net it s pronounced ree tee The Rete algorithm is imple mented by building a network of interconnected nodes Every node represents one or more tests found on the LHS of a rule Each node has one or two inputs and any number of outputs Facts that are being added to or removed from the working memory are processed by this network of nodes The input nodes are at the top of the network and the output nodes are at
178. a FALSE Boolean false a NIL No value NILLLIST An empty list a EOF End of file Many functions return TRUE when there is no other sensible return value Java s null is nota legal return value for call Jess s equivalent of nu11 is the symbol nil you can always therefore return Funcall NIL The function printout is one of the few built ins that return nil This value makes sense for printout because nil return values aren t displayed by the Jess command prompt Constructing Value objects Value objects are constructed by specifying the data and usually the type The type is one of the constants listed in table 15 1 Each overloaded constructor assures that the given data and the given type are compatible Note that for each constructor more than one value of the type parameter may be acceptable It s usually obvious which types can be created using each constructor for example you can only construct RU FLOAT objects using the constructor that takes a double argument The available constructors in jess Value are as follows public Value Object o throws JessException public Value String s int type throws JessException public Value Value v public Value ValueVector f int type throws JessException public Value double d int type throws JessException public Value int value int type throws JessException Value has a number of subclasses that are used to hold some specializ
179. a fact representing him using the duplicate function The duplicate function is similar to modify except that instead of modifying an existing fact it makes a copy and then modifies the copy Continu ing the previous example let s make a copy of the Jane Doe fact fact id 0 Jess gt duplicate 0 name John Doe gender Male lt Fact 1 gt Jess gt facts f 0 MAIN person name Jane Doe age 23 gender Female hobbies snowboarding restoring antiques f 1 MAIN person name John Doe age 23 gender Male hobbies snowboarding restoring antiques For a total of 2 facts Just like modify duplicate does nothing if the new fact would be an exact copy of an existing fact The duplicate function returns the fact id of the newly created 86 6 4 CHAPTER 6 Representing facts in Jess fact or FALSE if no fact was created The modify and duplicate functions only work with unordered facts because they require a slot name as an argument which ordered facts don t have Let s look at some things that ordered facts can do Ordered facts Although unordered slots are a great way to organize data when you need many slots per fact if you only need one slot they can seem redundant For example here s a deftemplate to hold a fact representing a single number Jess gt deftemplate number slot value TRUE Jess gt assert number value 123 lt Fact 0 gt The name value doesn t add any inform
180. a lang Class gt OBJECT lt External Address DimmerSwitchDynamic gt For a total of 1 facts Jess gt call ds setBrightness 10 Change brightness Jess gt facts property 94 6 5 8 6 6 CHAPTER 6 Representing facts in Jess f 0 MAIN dimmer Brightness slot updated brightness 10 to match Bean class lt External Address java lang Class gt OBJECT lt External Address DimmerSwitchDynamic gt For a total of 1 facts Shadow facts and working memory functions Many of the functions and constructs you ve seen so far work on shadow facts just as they work on other kinds of facts but there are some differences Most of these differences are obvious For example you can t use the assert function to assert a shadow fact you use definstance instead You can however use retract to remove a shadow fact You also can t put a shadow fact in a deffacts construct Note though that each individual definstance behaves like its own deffacts when you call reset the shadow fact for every instance is reinitialized The modify function operates on shadow facts just as it works on regular facts Furthermore if you modify the contents of a slot of a shadow fact Jess automati cally updates the corresponding JavaBean property This happens for both static and dynamic definstances Finally the duplicate function cannot be used with shadow facts Jess throws an exception if you try to duplicate one Summary Jess
181. a lot like the C functions with a few excep tions Some of the int types are now boolean because in Java boolean is the best type to represent an on or off value The heat pump state is still represented as an integer though and three constants represent the legal values which pre sumably came from the C library manual Writing a simulator The hardware simulator is fairly simple The implementations of methods like getVentState and setVentState are trivial The simulator object contains an array of boolean to record the state of each vent and these two methods simply set and get 226 CHAPTER 14 The reality connection the appropriate elements of that array The only complicated method is getTemperature because it must return a realistic temperature based on the com plete state of the system The simple part of the simulator is shown in listing 14 1 you ll work on getTemperature next Note that the implementation shown here like most published code skimps on error handling All the getter and setter methods should check that their first arguments are greater than zero and less than or equal to the num ber of devices but we ve omitted this code here to keep things short This version of Simulator is an abstract class because we haven t implemented get Temperature yet Listing 14 1 The basic parts of the hardware simulator package control public abstract class Simulator implements Hardware private int m_heatpumps pri
182. a rule The complete list of constructs includes defrule deftemplate defglobal deffacts deffunction and defquery they are all discussed in part 2 of this book Jess functions throw an exception of type jess JessException if the argu ments are of the wrong type lt number gt lt number gt Returns the first argument minus all subsequent arguments The return value is an INTEGER unless any of the arguments are FLOAT in which case it is a FLOAT lt number gt lt number gt Returns the first argument divided by all subsequent arguments The return value iS a FLOAT lt number gt lt number gt Returns the products of its arguments The return value is an INTEGER unless any of the arguments are FLOAT in which case it is a FLOAT Jess functions 391 lt number gt lt number gt Raises its first argument to the power of its second argument using Java s Math pow function The return value is NaN not a number if both arguments are negative lt number gt lt number gt Returns the sum of its arguments The return value is an INTEGER unless any of the arguments are FLOAT in which case it is a FLOAT lt lt number gt lt number gt Returns TRUE if each argument is less than the argument following it otherwise returns FALSE lt lt number gt lt number gt Returns TRUE if the value of each argument is less than or equal to the value of the argument f
183. adline t The first argument to printout or readline is a router a symbol that tells Jess where to send the output Several routers are built into Jess and they re all ini tially connected to standard input and output text sent to t WSTDOUT or WSTDERR all goes to System out by default and data read from t or WSTDIN comes from System in The W routers are used internally by Jess WSTDOUT is where Jess sends the Jess gt prompt and the result of evaluating an expression you type at the prompt WSTDERR is used for internal error messages WSTDIN exists only for sym metry Jess doesn t use it for anything A router is really just a symbolic name for a java io Reader for input or a java io Writer for output The defaults for the built in routers are wrappers around System in and System out Each jess Rete object keeps its own table of 320 18 6 1 CHAPTER 18 Embedding Jess in Java applications routers there are separate tables for input and output routers so an input out put pair with the same name like t are separate unconnected objects You can retrieve routers from this table by name with the getInputRouter and getOutputRouter methods If you re writing a Userfunction or other Java code that wants to intersperse its output with Jess s other output you can use getOut putRouter WSTDOUT to get an appropriate Writer object The method getOutStream is a special shortcut for this common operation Note th
184. agenda 19 20 22 97 121 210 211 391 air conditioner 257 algorithm 256 all products 301 alpha memory 138 alphabetical order 181 ambiguous method calls 69 analysis 24 and 392 conditional element 108 predicate function 106 Ant 324 328 330 433 Any CORBA type 241 Apache project 326 Apache Software License 326 appendText 321 applets 288 289 lifecycle issues 289 application object 393 application servers 222 288 324 355 359 apply 54 392 architecture 37 165 arguments type conversions 63 arrays 50 214 ART 32 136 asc 392 ask module 167 169 175 188 194 206 211 ask user 206 assert 78 211 392 assertFact 211 345 assertion storms 304 assert string 392 asynchronous events 264 ATOM 422 auto focus 127 128 169 191 303 304 automated testing See tests auto mated automobiles 189 availability 358 434 INDEX backward chaining 116 188 193 198 216 and queries 131 implementation 143 inhibiting 120 multilevel 119 preventing 202 backward chaining reactive 118 bag 392 BaseServlet 337 339 344 batch 47 162 259 393 batch files 428 BeanSupport 230 234 best match 69 beta memory 138 bias 157 binary decisions 269 bind 47 48 243 394 bit and 394 bit not 394 bit or 394 bload 235 247 394 boilerplate 291 books and journals 156 Boolean expressions 52 Boolean logic 264 Boolean values 43 BorderLayout 207 Bourne shell 428 breadth 409 bsave 235 247 394 buffer 257 build 55 394
185. ake each of these factors into account The current tem perature on each floor can be held in an array of double so that s easy Whether a given floor is being heated or cooled depends on what the corresponding heat pump is doing and whether that floor s vent is open or closed To account for heat leakage from outside you need a variable to hold the outside temperature Finally the same array of current temperatures is all you need to compute the heat leakage from other floors The simplest way to write the simulator is so that it works in real time it includes a loop that continuously recomputes the temperature This loop runs in a separate thread so the temperatures continue to update even if no calls to getTemperature are made Therefore all you need to do is to figure out an equation to calculate the current temperature from the temperature at the last time step and the changes due to the factors just listed A law of physics called Newton s Law of Cooling simply put states that the larger the temperature difference between two bodies the faster heat flows between them Therefore a reasonable way to calculate the temperature change per sec ond on a given floor is to calculate the difference between that floor s current temperature and some other body such as the outside air or the hot air coming from the heat pump and multiply this difference by some constant value giving the actual heat transfer rate The constant varies depend
186. al element The NoT node clearly can t function without a left input it can t create output tokens without left input tokens as seeds For this reason Jess inserts the initial fact pattern 8 6 2 8 6 3 8 7 8 7 1 Exploring the Rete network in Jess 143 into rules that have this problem Without this pattern the not conditional ele ment wouldn t work Implementing the test conditional element The test conditional element doesn t match a fact it simply evaluates a function To implement test Jess uses another special join node that ignores its right input For each token arriving on the left input the TEST join node evaluates its function If the function doesn t evaluate to FALSE the left input token and a null fact are composed to make a new token which is then sent to the node s output The TEST node just like the NoT node clearly can t work properly without a left input Therefore test conditional elements can also cause Jess to insert initial fact patterns into rules Implementing backward chaining Jess implements backward chaining on top of the forward chaining Rete algo rithm The basic problem is detecting when the LHS of a rule has been matched up to but not including a pattern that matches a backward chaining reactive tem plate see section 7 4 At this point a signal must be sent describing the fact that would complete the match as you might guess having read section 7 4 that sig nal
187. aldo Pinali Doederlein Pau Ortega Peter Hanson Peter Klotz Ralph Grove Richard Long Rob Jefson Robert Gaimari Russ Milliken S S Ozsariyildiz Sander Faas Scott Kaplan Scott Track man Sebastian Varges Seung Lee Sidney Bailin Simon Blackwell Simon Hamil ton Steve Bucuvalas Thomas Barnekow Thomas Gentsch Travis Nelson William E Wheeler Win Carus and Yang Xiao I m sure I ve forgotten someone important please forgive the oversight The staff at Manning Publications both past and present are talented people and real professionals I thank Marjan Bace for his guidance and eye for the big picture Lianna Wlasiuk for her useful and practical advice in the first stages of writing Ann Navarro for her expertise in editing Tiffany Taylor for the tremen dous skill and effort she applied to meticulously copy editing and formatting my ill formed manuscript Syd Brown who produced the beautiful example of the typographer s art you see before you Maggie Mitchell for proofreading Mary Piergies for overseeing the production of this book Ted Kennedy for gathering a team of excellent reviewers and organizing the results Dan Barthel who got me started on this book in the first place and undoubtedly many others who worked behind the scenes Quite a few technical reviewers and friends read the manuscript and provided detailed and useful comments This book is vastly improved by their input any remaining problems are of course
188. alling run is enough to do it Behind that call to run all the rules you developed in chapter 17 get a chance to fire When they do they create recommend facts and the query recommendations for order retrieves them engine run Iterator result engine runQuery recommendations for order new ValueVector add orderNumberValue 346 CHAPTER 19 Deploying web based applications NOTE A note about calling run here Your web application might have multiple users at the same time each running doGet in a separate thread Howev er each user has a unique customer ID order number pair so the Rec ommendations Agent can keep track of them separately Sometimes one thread might call run resulting in rules being fired in response to facts asserted from another thread but this is OK Other times one thread might call run right after another thread already has done so and this second call to run will have no effect this is also OK As long as the rules aren t specifically written to expect run to be called only from certain states everything will work fine Jess is meant to be used in multithreaded environments and it uses synchronization to protect potentially fragile code from being corrupted by concurrent access 19 10 4 Forwarding to JSPs 19 11 If there are recommendations you want to give the customer the chance to pur chase the recommended products If not you can go straight to the purchase page to get shippin
189. alls Let me say that again The left hand side of a rule the if part consists of pat terns that match facts they are not function calls The right hand side of a rule the then clause is made up of function calls The following rule does not work Jess gt defrule wrong rule eq 1 1 gt printout t Just as I thought 1 1 crlf Many novice Jess users write rules like this intending eq 1 1 to be interpreted as a function call This rule will not fire just because the function call eq 1 1 would evaluate to TRUE Instead Jess tries to find a fact in the working memory that looks like eq 1 1 Unless you have previously asserted such a fact this rule will not be activated and will not fire If you want to fire a rule based on the evaluation of a function you can use the test conditional element described in section 7 3 4 Forward chaining rules 99 The example rule then will be activated when the fact baby is wet appears in the working memory When the rule fires the function change baby is called and the baby is wet fact is retracted This rule forms part of a com plete program in listing 7 1 Jess gt clear TRUE Jess gt watch all TRUE Jess gt reset gt f 0 MAIN initial fact TRUE Jess gt deffunction change baby printout t Baby is now dry crlf TRUE Jess gt defrule change baby if wet If baby is wet change its diaper wet lt baby is wet gt change baby retract
190. alue of the property and the set method to change it The java beans API includes a class named Introspector that can 1 Plenty of great books on JavaBeans are available if you re interested For example Lawrence Rodrigues The Awesome Power of Java Beans Greenwich CT Manning Publications 1998 Shadow facts 89 i slot propertyTwo E a public String getPropertyOne public void setPropertyOne String public int getPropertyTwo public void setPropertyTwo int a Jess deftemplate JavaBean slot propertyOne public class JavaBean Figure 6 3 Jess uses the java beans Introspector class to turn JavaBeans into deftemplates examine JavaBeans and find properties defined according to this get set nam ing system There are a few wrinkles having to do with capitalization and other details but this simple convention works most of the time Jess can use Introspector to automatically generate a deftemplate that repre sents any specific JavaBean class see figure 6 3 A fact created from this deftemplate can then serve as a sort of adapter to store the JavaBean in working memory the shadow facts A shadow fact has one slot for each JavaBean property If a JavaBean has array properties those properties become multislots and all other properties become normal slots The slots are automatically populated with the values of the JavaBean s properties In the next section you ll watch Jess use Introspec
191. an be valuable although I personally find that I use the HttpUnit approach more often TIP One word of advice before you begin writing and testing a large applica tion like this you ll want to first carefully test and debug your rules as much as possible using Jess s command line interface The Recommend servlet Recall that when the login form is submitted the Catalog servlet gets a single request parameter named customerId When the catalog form is submitted there is an items request parameter for each product the customer purchases and that will serve as the input to the Recommend servlet The customerId and orderNumber will still be in the HttpSession and the Rete object will be already initialized and stored in the ServletContext The Recommend servlet has several tasks to do 1 Gather the items customerId and orderNumber 2 Make sure the customer is logged in and is ordering something 3 Reset the order in case the form is resubmitted 4 Create an order fact 344 19 10 1 19 10 2 CHAPTER 19 Deploying web based applications 5 Create a series of line item facts 6 Get recommendations from Jess 7 Pass the recommendations on to a JSP for display This servlet has to use more of Jess s Java APIs than the Catalog servlet does Let s go through these steps in order as we walk though the Recommend source Getting started The Recommend servlet extends the BaseServlet class It calls checkInitialized
192. ance for later cycles is the same as well This means Jess won t be very efficient if you populate the working memory run the pattern matcher for just one cycle and then reset working memory and repopulate it again from scratch But in the typical case of a slowly changing work ing memory moderate sharing in the network and effective indexing Rete will vastly outperform the naive algorithm for all cycles after the first The runtime will be proportional to something like R F where R is a number less than R the number of rules F is the number of facts that change on each iteration and P is a number greater than one but less than the average number of patterns per rule Node index hash value Jess uses a fairly sophisticated data structure to represent the two memories in each join node It is basically a hash table with a fixed number of buckets The interesting feature here is that the hash function uses the part of each token that is most relevant to the specific tests in each join node This lets Jess presort the tokens before running the join node tests In many cases running some or all of the tests can be avoided altogether because once the tokens are sorted into the hash buckets questions involving comparisons of slot data can often be answered categorically for a whole bin The node index hash value is a tunable performance related parameter that can be set globally or on a per rule basis It is simply the number of buc
193. and As soon as you enter run the activated rule fires Because you entered the watch all com mand Jess prints the diagnostic FIRE 1 MAIN change baby if wet f 1 to notify you of this action The f 1 is a list of the facts that matched this rule s LHS You then see the output of the rule s RHS actions First the function change baby is called Second the fact f 1 is retracted The variable wet is called a pat tern binding the lt operator stores a reference to the fact baby is wet in this variable and the retract function can then access this variable on the rule s RHS Note then that rules cannot only react to the contents of working mem ory they can change it Thus one rule can put information into working mem ory which in turn can cause another rule to fire The final number 1 is the number of rules that fired the return value of the run command The run function returns when there are no more activated rules to fire What would happen if you entered run again Nothing Jess activates a rule only once for a given working memory state Once the rule has fired it will not fire again for the same list of facts You won t change the baby again until a new baby is wet fact is asserted Patterns and shadow facts Jess s working memory can hold JavaBeans as well as facts Actually you ll recall that this isn t quite correct The working memory contains only facts but some of those facts called shadow facts a
194. and design data structures that will make it easy to implement the rules clearly and directly This process resembles object oriented analysis First the major concepts are identified For an employee benefits consultant these might include employee health plan claim time and money The important thing at this stage is to identify all the concepts referred to in the collected knowledge the irrelevant ones can be removed later Then you list all the variable characteristics of each concept Employees have a name a health plan years of service and a salary among other things Again at this stage you try to identify all the characteristics mentioned in the collected knowledge The pants color and position templates in chapter 1 were simple examples of data structures for working memory elements Designing data struc tures for rule based systems is discussed in chapter 10 Testing You may wonder why I m mentioning testing now when you haven t written any code yet Actually this is the perfect time to begin testing a rule based system at the beginning If rigorous tests are applied to the system at every stage of its development it will naturally be more robust more modular and better under stood than a system that wasn t tested until the end Therefore before writing a group of rules you should develop an automated test to exercise them You can write tests in Java in your rule language or in a convenient scripting language
195. and later items For small collec tions a selection sort is a perfectly reasonable choice and you ll use it here This rule based implementation of a selection sort compares the names of forms in recommendation facts The single rule prints information about a fact and then retracts it if and only if there are no other recommendation facts with a form that comes alphabetically earlier Each time you retract a fact a different recommendation becomes the new alphabetically first one until all the recommendation facts are exhausted You can use the str compare function which takes two strings as arguments and returns 1 if the first one is alphabetically first 1 if the second string should come first and 0 if the arguments are equal defmodule report defrule sort and print r1 lt recommendation form f1 explanation e not recommendation form f 2 amp lt str compare 2 f1 0 gt printout t Please take a copy of form f1 crlf printout t Explanation e crlf crlf retract r1 182 11 4 1 11 5 CHAPTER 11 Writing the application Testing the report module The report module like the startup module contains only one rule You can test it by asserting some recommendation facts yourself form f3 explanation ef3 explanation ef1 explanation ef2 explanation ef4 Jess gt assert recommendation recommendation form f1 recommendation form f2 recommendation form f
196. and two functions and you can call it just by asserting a fact Summary In this chapter you began to turn the knowledge you developed in chapter 9 into a concrete rule based system You determined the form the data in working mem Summary 171 ory will take you partitioned the system into modules that represent the phases of processing and you developed some input output functionality you will use later In the next chapter you will develop the rules that form the actual application We ll pay special attention to testing techniques so that you ll trust the compo nents of the application to work well individually and as a complete system Writing the application In this chapter you ll a Write the rules for the Tax Forms Advisor a Work with multiple modules Test the completed application 172 11 1 11 1 1 Welcoming the user 173 You ve done the research laid the groundwork and sketched out a design so now it s time to write the rules that implement the Tax Forms Advisor application In this chapter you ll write the rules test them and then test the completed application As you ll recall from chapter 10 the completed application will consist of five modules ask startup interview recommend and report You developed the ask module in chapter 10 In this chapter you ll write the rules for the other four modules one module at a time The code for the complete application is available from this bo
197. and visit the URL http 127 0 0 1 8080 Hello servlet Hello to see your first servlet in action The result is shown in figure 19 1 pe Hello World Figure 19 1 The result of viewing the Hello servlet in Netscape 19 3 Your first Jess servlet Now that you ve built a simple servlet it s time to begin developing the Recom mendations Agent web application The first small step is to write a servlet that embeds Jess however simply Why not modify Hello to create HelloJess which uses printout to create the same simple web page import import import import public jess java io javax servlet javax servlet http class HelloJess extends HttpServlet public void doGet HttpServletRequest request HttpServletResponse response throws IOException ServletException response setContentType text html PrintWriter out response getWriter Rete engine new Rete engine addOutputRouter page out 330 19 3 1 CHAPTER 19 Deploying web based applications try print lt html gt engine print lt head gt engine print lt title gt Hello World lt title gt engine print lt head gt engine print lt body gt engine print lt h1 gt Hello World from Jess lt h1 gt engine print lt body gt engine print lt html gt engine catch JessException je throw new ServletException je private void
198. any Userfunction can take a variable number of arguments unless you take specific action to enforce a certain length for your argument list If your 15 2 2 Handling arguments 241 Userfunction accepts precisely one argument you might begin your implemen tation of call like this public Value call ValueVector vv Context context throws JessException if vv size 2 throw new JessException kill Wrong number of arguments vv size 1 On the other hand a function that takes an arbitrary list of arguments could use size as the upper limit of a processing loop Using arguments Suppose that the one argument to ki11 is the value that should be passed to Sys tem exit You might write the following public Value call ValueVector vv Context context if vv size 2 throw new JessException kill Wrong number of arguments vv size 1 System exit vv get 1 intValue context return null The values returned by ValueVector get are always jess Value objects A jess Value is a kind of generic wrapper for any Jess data object something like a VARIANT in Microsoft COM programming or an Any in CORBA programming Once it is constructed a Value s type and contents cannot be changed it is immu table You might know that Java defines many such immutable classes Integer and String are two well known examples Value has a type member function you can call to find out what kind o
199. ar x void println char x void println double x void println float x void println int x void println long x void println Object x void printlin String x When you call an overloaded method in Java code the Java compiler chooses an overload based on the exact compile time types of the parameters Java chooses the 5 3 Accessing Java member data 69 most specific applicable method Sometimes if there s no clear choice you get a Java compiler error but the important point is that there s always a right answer Jess is much more relaxed about choosing between overloads because it has to be It doesn t have the same kind of strict type information that Java has One sim ple example Looking at the list of overloaded print1n methods you can see that there are versions for both double and float Jess has only one floating point type so it can t be sure which one you d rather call When you call an overloaded method such as print1n Jess looks at each of the overloads in turn trying to match the parameter types of the method to the types of the arguments you passed The first overload Jess finds that can be invoked with the given argument list will be called Jess does not search for a best match it uses the first matching method it finds Because Jess knows so many different ways to convert between Jess and Java values the whole idea of a best match is too vague to be useful Often it doesn t matter which of a s
200. aram name gt rulesfile lt param name gt lt param value gt tekmart clp lt param value gt lt context param gt lt context param gt lt param name gt factsfile lt param name gt lt param value gt var tekmart data lt param value gt lt context param gt These are the two parameters read by the checkInitialized method var tek mart data is the absolute path to a file data where the order information should be persisted What s next This is a book about rule based application development not a book about build ing e commerce systems so the web application you ve developed here is lacking in many important respects Security for example is nonexistent and of course the system doesn t do anything with the order the customer eventually places These kinds of enhancements are straightforward though and many other books discuss them There s no point in going into them here On the other hand there are some ways in which the use of Jess in this applica tion could be improved You could encapsulate all the knowledge about Jess in the servlets themselves so the JSPs don t depend on the specific rule engine being used The JSPs work with Iterator objects passed in as HttpServletRequest attributes Instead of passing the iterators returned from Jess the servlets could construct neutral data structures using only generic containers of strings and 19 16 Summary 353 numbers and pass iterators over these ne
201. are named startup interview recommend and report respectively Defmodules partition not only the rules of an application but also the facts You need to decide which of the templates ought to go into which of the modules You can do this by looking at which module s rules need access to the data Remember that if two or more modules need to share a deftemplate it should go into the module MAIN Examination of the list of templates and of the modules listed here shows that every template will be needed by at least two modules For instance the question and answer templates need to be shared between the interview and recommend modules whereas recommendation is needed by both recommend which asserts recommendation facts and by report which displays information derived from them As a result all of the deftemplates you define will be in module MAIN This is not unusual Building the infrastructure Very often many of the rules in a rule based system follow a repeating pattern You know this application needs to ask the user a series of questions and record the 10 6 1 Building the infrastructure 167 answers in the working memory You can develop code to ask a question and receive an answer as a kind of subroutine and all the rules that need this capability can call it Not only does this approach simplify the code for your system but it also makes it easier to change the interface if you need to upgrade from a text based to a
202. argeted toward this product niche Part VII of this book discusses the use of rule engines in general and the Jess rule engine in particular in enterprise applications based on the Java 2 Enterprise Edi tion J2EE architecture Summary Rule based programs are everywhere Their applications include everything from mail filtering to order configuration and from monitoring chemical plants to diagnosing medical problems Rule based programs excel at solving problems that are difficult to solve using traditional algorithmic methods and they work even when the input data is incomplete In the next chapter we ll refine our definition of a rule based system You ll learn about the history of rule based systems and about their architecture We ll also look at the development cycle for rule based programs By the end of the next chapter you ll be ready to begin learning how to write your own rule based software What are rule based systems In this chapter you ll Be introduced to declarative programming Learn the architecture of a typical rule based system See a method for developing rule based systems Read about industry standards for rules and rule engine APIs 13 14 2 1 CHAPTER 2 What are rule based systems There s an old saying that when all you ve got is a hammer everything looks like a nail In computer programming too it s important to choose the right tool for the right job Some problems
203. arguments but your video game will contain scenes with dozens of rival ninjas A version of min that takes an unlimited number of arguments would be very useful and you can write such a function in Jess In the definition of a deffunction the last parameter can be a multifield a variable starting with as in x If it is then the deffunction will accept a variable number of arguments any arguments in excess of the number of named parameters are compiled into a list and that list is assigned to the multi field Here s a version of min that uses this feature Jess gt deffunction min args Compute the smallest of a list of positive numbers bind minval nths 1 args foreach n Pargs if lt n minval then bind minval n return minval TRUE Jess gt min 10 100 77 6 43 6 Late binding Jess uses late binding for function names meaning the link between the name of a function and the code it represents can be changed right up until the instant the function is called In Java of course code won t compile if it calls a function that hasn t been defined yet In Jess you can write code that calls a function even if you haven t written that second function yet Jess won t try to find the undefined function until the code that calls it is executed If the undefined function hasn t been defined at that time an error will result Late binding is a useful property because it also means you can redefin
204. as lists where the head is the special symbol defrule Lists remind many new Jess programmers of scopes in Java the parentheses group the items in a Jess list just as pairs of brackets delimit Java methods and classes As a result some people tend to write Jess code by lining up opening and closing parentheses vertically as they might do in Java In the worst cases that might look like this bind x In addition to being aesthetically awkward it can be confusing It s better style to simply put the closing parentheses on the last line of the list they close like this bind x 20 3 37 23 46 4 2 2 CHAPTER 4 Getting started with the Jess language This is generally more readable not less Whether you use a text editor that helps you balance parentheses may influence your choice of code format The open source Emacs editor is available on the Windows Linux and Macintosh plat forms and makes editing Jess code easy A special Jess mode for Emacs is available http jess mode sourceforge net Jess interprets lists in many different ways depending on where in your pro gram they appear A list might be a rule a fact a pattern on a rule s left hand side a function definition or a function call among other things You ll see ways to use lists as this chapter continues but let s pick just one to get started The simplest Jess program consists of a single function call and you re probably anxious
205. asks are part of application deployment Eventu ally javax rules compliant application servers should offer dedicated mecha nisms GUIs for installing a rule engine and preparing rule sessions just as they currently include tools for plugging in databases and other resources The JSR 94 reference implementation includes a JCA connector that makes a RuleService Provider accessible through JNDI Setting things up The administrative phase begins with finding an appropriate javax rules Rule ServiceProvider object which gives the programmer access to the rest of the javax rules implementation In a J2EE environment you may be able to retrieve the RuleServiceProvider using JNDI Otherwise you can get one from the javax rules RuleServiceProviderManager class String implName org jcp jsr94 ri RuleServiceProvider Class forName imp1Name RuleServiceProvider serviceProvider RuleServiceProviderManager getRuleServiceProvider imp1Name JSR 94 the javax rules API 385 This is a little like loading a JDBC driver it s a similar concept in that both pro vide access to an external resource and you need to do some general initializa tion before you instantiate any of the classes Once you have the RuleServiceProvider you can retrieve a javax rules admin RuleAdministrator From the RuleAdministrator you can get a RuleExecu tionSetProvider which as the name suggests creates javax rules RuleExecu tionSets A RuleExecutionS
206. at Jess s routers are not general I O channels Jess always interprets data from an input router as a sequence of Jess language tokens symbols num bers and strings As a result you can t do binary I O through a router If you need to do binary I O you can use Jess s reflection capabilities to call read and write directly on appropriate Java streams Using custom routers In the command line client jess Main the default values for the built in routers are fine but if Jess is embedded in software with a GUI it s likely that printing to standard output is inappropriate If you want to be able to use readline print out and friends from an application that won t be used from the command line you may need to set the standard routers to refer to Reader and Writer objects that are more appropriate for the situation For example if Jess was to be embed ded in a web application printout wouldn t be useful unless a new router were defined that sent its output to someplace other than System out You might define a new router using a java io StringWriter printout could then send information to a string which could then be retrieved when appropriate by the application To define such a router named out you could do the following StringWriter sw new StringWriter Engine addOutputRouter out sw If you then executed the Jess statement printout out 12345 crlf a subsequent call to sw toString would return the string 12345
207. ates any fractional component of the value of the given numeric expression and returns the integral part integerp lt expression gt Returns TRUE for INTEGER values otherwise returns FALSE intersection lt list gt lt list gt Returns the intersection of two lists a list consisting of the elements the two argu ment lists have in common jess version number Returns a version number for Jess currently 6 1 jess version string Returns a human readable string descriptive of this version of Jess length lt list gt Returns the number of fields in a list value lexemep lt expression gt Returns TRUE for symbols and strings otherwise returns FALSE list deftemplates module name With no arguments prints a list of all deftemplates in the current module not the focus module to the t router With a module name for an argument prints the names of the templates in that module With as an argument prints the names of all templates list focus stack Displays the module focus stack one module per line the top of the stack the focus module is displayed first list function Returns a list of all the functions currently callable including intrinsics deffunctions and Java user functions Each function name is a symbol The list is sorted in alphabetical order load facts lt lexeme gt Asserts facts loaded from a file The argument should name a file containing a list of facts not deffacts
208. atically whenever the JavaBean s properties change The definstance you have defined already is static What happens if you change the brightness property of your Bean turning on the light You can invoke the setBrightness method using the call function you learned about in chapter 5 Jess gt call ds setBrightness 10 Jess gt facts f 0 MAIN dimmer brightness 0 class lt External Address java lang Class gt OBJECT lt External Address DimmerSwitch gt For a total of 1 facts Nothing changed The working memory still thinks the Bean s brightness is set at 0 even though you changed it to 10 This is expected behavior for a static definstance Obviously though you d like the shadow fact to track changes in the Bean s properties What happens if you call reset 92 6 5 7 CHAPTER 6 Representing facts in Jess Jess gt reset TRUE Jess gt facts f 0 MAIN initial fact f 1 MAIN simple brightness 10 class lt External Address java lang Class gt OBJECT lt External Address DimmerSwitch gt For a total of 2 facts The reset function updates the shadow fact to match the JavaBean This behavior is what you get when you specify static in the definstance function as you did earlier Static shadow facts are refreshed only when a reset is issued In between reset calls their properties do not change in response to property changes in their JavaBean If you want to have a shadow fact continuo
209. ation it just clutters things up For sim ple cases like this it would be nice to be able to omit the notion of a slot name altogether and Jess allows this Jess gt clear TRUE Jess gt assert number 123 lt Fact 0 gt You can assert facts that look like simple flat lists without explicitly defining a deftemplate for them as long as no deftemplate using that same head has already been defined All the facts you used in the first section of this chapter were ordered facts We ll use the following functions to take a closer look at ordered facts and their templates ppdeftemplate Displays a pretty printed deftemplate m show deftemplates Lists all the deftemplates currently defined When you assert the first ordered fact with a given head Jess automatically gener ates an implied deftemplate for it The ppdeftemplate function lets you see the definition of any deftemplate Let s use it to examine the implied deftemplate for number Jess gt ppdeftemplate number deftemplate MAIN number extends MAIN __fact Implied multislot __data podeftemplate the pp stands for pretty print returns its result as a string so the definition is enclosed in double quotes and the quotes around the documentation string are escaped with backslashes The implied deftemplate number contains a 6 5 Shadow facts 87 single slot named __ data Jess treats facts created from these deftemplates spe cially The name of
210. ation form 8829 explanation ex recommendation form 1040 explanation ex Explaining the results 181 11 4 Explaining the results When the recommend module is finished the working memory includes a series of recommendation facts representing the application s suggestions for the user All the report module needs to do is display the data from these facts in a sensi ble way In particular the forms should be listed in alphabetical order note that then the main forms 1040 1040A 1040EZ are displayed first Each recommendation fact includes a bit of text rationalizing the recommen dation in the explanation slot You ll display these explanations too Because the explanations can span multiple lines you can t easily display the program s output in tabular form at least not if you have to stick to plain text If you could use HTML the display would be easy You ll be developing web based systems later in the book so for simplicity s sake you ll display each recommended form and its explanation together one on top of the other Sorting data is an old and well studied topic and there are many ways to accom plish a sorting task A simple if not particularly efficient technique is the selection sort To perform a selection sort you search through a collection for the smallest or largest item and then remove it and make it the first item in a new sorted col lection Repeat this process to find the second
211. ation for a Java enterprise application architecture J2SE Java 2 Standard Edition The main Java software developer s kit JCA J2EE Connector Architecture A generic API for accessing resources from EJBs JDBC Java Database Connectivity The standard Java API for working with relational databases JNDI Java Naming and Directory Interface The standard Java naming service in the J2EE environment JSR94 The javax rules API A developing standard API for rule engines LDAP Lightweight Directory Access Protocol A standard enterprise naming service RMI Remote Method Invocation A Java API for connecting Java applications over a network XML Extensible Markup Language A self describing structured text document format XSLT Extensible Style Language Transformations A declarative pattern language for transforming XML documents Just as J2EE is a universal software architecture for building enterprise applica tions XML can be seen as a universal data structuring architecture for the enter prise To use Jess in the enterprise you ll want to embed it in a J2EE application likewise you ll want to represent the rules themselves in XML Rules and XML A number of projects exist with the aim of defining a standard rule language based on XML the Extensible Markup Language XML is a self describing text based structured data file format The power of XML lies not so much with XML itself but with the wealth o
212. ava Map for its collection of keys and you can ask that collection to convert itself to an array If you could convert that array to a plain list in Jess you d be able to recre ate the simple grocery list you worked with in earlier chapters of this book As noted in tables 5 1 and 5 2 Jess automatically converts Java arrays to plain lists Values of type RU LIST You can use the method toArray in java util Collection to extract all the keys from your HashMap into a Jess list Jess gt bind grocery list prices keySet toArray bread peas beans If you want to put your list of groceries into a pop up menu component you can pass this list of items as a constructor argument to the javax swing JComboBox class JComboBox wants an array as a constructor argument but Jess converts your plain list back into an array automatically Jess gt import javax swing JComboBox Jess gt bind jcb new JComboBox grocery 1list lt External Address javax swing JComboBox gt 68 5 2 5 CHAPTER 5 Scripting Java with Jess This system works well for small arrays less than a few dozen items but convert ing between arrays and lists is inefficient for large arrays because the Jess data structures to represent the plain list must be created or destroyed on each conver sion A better way of working with large Java arrays is planned for a future version of Jess Meanwhile if you need to work with large arrays in your Jess programs you can
213. ax forms the user might need The list doesn t need to be exhaustive but it should be generous that is if in doubt recommend the form The series of questions should be as short as possible and should never include irrelevant or redundant questions This simple statement is certainly enough to get you started on the knowledge engineering phase of the project An interview If you ve ever filed your own income taxes you have a reasonable understanding of the concepts behind this application So you probably don t need to do any advance desk work in fact you can probably gather all the necessary information 158 CHAPTER 9 Collecting the knowledge from one or two interviews and from reading the forms themselves The first step is to talk to an accountant and ask her to list the 10 most used income tax forms She gives you this list in no particular order Form 1040 Income tax Form 1040A Income tax Form 1040EZ Income tax Form 2441 Child and dependent care expenses Form 2016EZ Employee business expenses Form 3903 Moving expenses Form 4684 Casualties and thefts Form 4868 Application for filing extension Form 8283 Noncash charitable distributions Form S829 Home office expenses With the list in hand you can begin asking questions about individual forms The most relevant question for each form is Who needs it The accountant s answers are reproduced here Form 1040 is the standard
214. ay use a Microsoft operating system or the Jakarta project s Ant build man ager or you may have a commercial J2EE implementation That s fine but you may need to adjust file paths and command lines to suit your environment I assume you understand your tools well enough to do this The Java Servlet API 325 19 1 The Java Serviet API You re going to build this application using the Java Servlet API http java sun com products servlet index html We talked briefly about servlets in chapter 17 A servlet is a bit like a device driver It s a chunk of code that imple ments a service and provides a small number of entry points through which the host system can access that service A servlet is of course written in Java and the most common variety of servlet is hosted by a web server and accessed via the web using the HTTP or HTTPS protocol Here s a simple but complete servlet import java io import javax servlet http public class Hello extends HttpServlet public void doGet HttpServletRequest request HttpServletResponse response throws IOException response setContentType text html PrintWriter out response getWriter out println lt html gt out println lt head gt out printlin lt title gt Hello World lt title gt out println lt head gt out println lt body gt out printin lt h1 gt Hello World lt h1 gt out println lt body gt out println
215. before advice this means the original function is not called defclass lt lexeme gt lt lexeme gt extends lt lexeme gt Defines a template based on a Java class The first argument is the template tag and the second argument is the name of a Java class The names of the template s slots are based on the JavaBeans properties found in the named class If the optional extends clause is included the last argument is the name of another template it becomes the parent template of the new template The common slots in the two templates are in the same order at the beginning of the new template Jess functions 397 Rules defined to match instances of the parent template also match instances of the new child template definstance lt lexeme gt lt object gt static dynamic Creates a shadow fact representing the given Java object according to the tem plate named by the first argument which should have come from defclass If the symbol static is not supplied as the optional third argument a PropertyChangeListener is installed in the given object so that Jess can keep the shadow fact updated if the object s properties change dynamic is the default The shadow fact is returned Note that it is an error for a given Java object to be installed in more than one definstance at a time The second and subsequent definstance calls for a given object will return a fact with an ID of 1 delete lt list gt lt integer gt lt inte
216. ble describe the desired output of the system given those inputs A vastly simplified truth table for the HVAC Controller system might look like table 16 1 In the following sec tions you ll fill in the details 16 1 2 16 1 3 The control algorithm 257 Table 16 1 Oversimplified truth table for the HVAC Controller system Each row represents one pos sible current heat pump setting each column represents one possible temperature reading Too Cold Just Right Too Hot Off Start Heating Do Nothing Start Cooling Heating Do Nothing Turn Off Turn Off Cooling Turn Off Turn Off Do Nothing How heat pumps work You may know how an air conditioner works A working fluid historically often the gas Freon is compressed by a pump outside your house Compressing a fluid cre ates heat so a fan dissipates this heat outside The compressed somewhat cooled fluid is then circulated inside the house where it is allowed to expand quickly Just as compressing a fluid creates heat an expanding gas absorbs heat This process cools a radiator inside the house and a fan blows through the radiator to create a nice cool breeze The expanded fluid is pumped back outside to be compressed again A heat pump is basically an air conditioner that can run in either direction It can either cool the inside of a building and expel hot air outside or cool the out side air and expel the waste heat inside heating the building Using guard lin
217. ble if it wants to return a different value 60 4 5 1 4 6 CHAPTER 4 Getting started with the Jess language Take my advice please You can only apply advice to actual functions built in or user defined not to function like constructs like deffunction defglobal deftemplate defrule or defquery Here s a short deffunction to confirm whether something is a function Jess gt deffunction is a function name neq FALSE member name list function TRUE Jess gt is a function printout TRUE Jess gt is a function deftemplate FALSE Let s figure out how this function works it ll be good practice in reading Jess code and it includes calls to a couple of built in functions you haven t seen before To read complex nested Jess language expressions it s usually best to start from the inside and work your way out The innermost function call in this deffunction is to list function s which returns not surprisingly a list of all the functions cur rently defined in Jess including all deffunctions and user defined functions writ ten in Java The member function returns FALSE if the first argument is not found in the second argument which must be a list and returns the index at which the first argument was found otherwise Finally is a function uses neq Not EQual to convert member s somewhat odd return value into a simple TRUE or FALSE The whole def function then returns TRUE if the argument app
218. ble to be read from the Text Reader s internal buffer The connected TextReader object is installed as the t and WSTDIN input routers Summary Jess is designed as a library that can be embedded into many different kinds of software It has an extensive Java API to help you accomplish this functionality and we have covered only some of it in this chapter The jess Rete class plays a central role Its methods let you define and retrieve constructs like rules and templates You can use store and fetch to pass values between Jess and Java code jess Rete also manages a set of I O routers that control how Jess prints output and reads input You can work with jess Fact objects from Java whenever you need more con trol over the creation or manipulation of facts than you can get from the Jess lan guage Working with facts often involves manipulating the jess Deftemplate objects that define the structure of the facts You can also add JavaBeans to Jess s working memory from Java code using the defclass and definstance methods of the Rete class 322 CHAPTER 18 Embedding Jess in Java applications Jess uses the jess JessException class to report errors and you can use it to report errors when you write Userfunctions A JessException often carries a wealth of useful information about the error it represents Jess s I O functions read input from and write input to objects called routers You can install your own routers redirect the pred
219. blic void setSlotValue String name Value value Sets the value in a slot by name Don t call this method on a Fact that is currently in Jess s working memory B 3 jess Funcall A Funcall is a specialized list representing a function call If you need to calla Jess function from Java repeatedly with no or only minor changes it s generally more efficient to build a Funcall object yourself and call execute on it Funcall also includes a number of useful constants as static Value objects Funcall TRUE Funcall FALSE and Funcall NIL are all symbols and Funcall NILLIST is an empty list public Funcall String name Rete engine Accepts the name of the function to call Note that the function doesn t have to be defined at the time as long as it is defined when execute is called everything will work public Funcall arg Value v Appends an argument to the Funcall Note that because this method returns the modified Funcall object you can chain calls to arg together public Value execute Context context Executes the Funcall1 using the given Context If you don t have a Context handy you can get one by calling Rete getGlobalContext B 4 jess Jesp Jesp is the Jess parser You might use this class if you wanted Jess to read input from a network stream or some other unusual source public Jesp java io Reader reader Rete engine Takes a Reader object the source for all the input this object will parse The stream of Je
220. build xml 330 business rules 11 355 362 buzzwords 358 360 c C 224 225 call 64 317 395 Jess function 215 omitting 65 Userfunction method 239 call on engine 395 categories 293 CD ROMs 157 checklInitialized 337 344 child care expenses 176 clarity 297 class variables 70 classical set theory 266 classification 265 CLASSPATH 33 232 clear 79 210 311 395 clear focus stack 127 395 clear storage 395 client server 288 CLIPS 32 85 136 performance 38 close 395 CLP files 162 code as data 54 Collections API 62 67 COM 241 combo box 213 command line interface 150 206 comments 44 152 nested 44 complement 396 complete testing 177 component architecture 376 components 88 222 conclusions 17 75 concurrency 302 condensation 257 conditional elements 108 and 108 combining 112 exists 113 explicit 120 202 logical 115 not 111 142 296 or 109 ordering for efficiency 111 reading 112 subrules 110 test 113 143 198 299 conflict resolution 20 121 breadth 121 depth 121 strategy 121 201 conflict set 20 121 conflict strategy 22 conflicting information 154 constants 233 constraints 101 blank variables 104 connective 104 297 functions 105 global variables 104 grouping 105 literal 101 multifields 103 pattern bindings 107 precedence 105 return value 106 variables 102 constructs 60 390 container 376 context 396 Context object 416 control 225 control algorithm 256 control engineering
221. c Assistant Please wait Figure 13 3 Application window with both an area for displaying questions and a lower panel for collecting answers 13 4 1 Getting answers 209 a actionPerformed ActionEvent event jess awt ActionListener Figure 13 4 An instance of the class jess awt ActionListener that serves as an event adapter and forwards GUI events to the deffunction read input deffunction read input EVENT read input event the event adapter invokes the deffunction passing it the event object as the lone argument the parameter EVENT in the listing This arrangement is shown in figure 13 4 Now you have an input field that asserts a Jess fact in response to a button click or the user pressing Enter Before you can modify the program to use this asserted fact however we need to look in a bit more depth at the architecture of a GUI based application The main thread vs the event thread A thread is a single flow of control in a Java program Individual streams of code can be and often are running in multiple separate threads simultaneously You can create threads explicitly using the java lang Thread class and Java creates some threads on its own for example the main thread from which the main function is called Every Java program is a multithreaded program Even in a Hello World pro gram where the user s code clearly runs only in the main thread the JVM creates a number o
222. can be solved easily using traditional pro gramming techniques whereas writing a rule based system is the easiest way to solve others Other problems are in the middle and either technique will work equally well This chapter will help you understand which problems are well suited to being solved with a rule based system and what this type of software is all about The cooking driving robot Imagine your first day on the job at Acme Advanced Robotics Corporation Acme has built some great humanoid robot prototypes but the company is hav ing trouble teaching its robots to survive in the real world You ve been brought in to write some high level robot control software so that Acme can begin selling robot butlers to the masses You decide to start with something simple breakfast How do you teach a robot to prepare a bowl of breakfast cereal Assume for the moment that the robot knows how to recognize pick up and interact with objects and that the robot will operate in a closed controlled optimal environ ment It is be a straightforward program Tell the robot to get a bowl a spoon and a napkin and set them on the table Then the robot should open a box of cereal fill the bowl and add milk If you were writing a computer program for a sophisticated cereal serving robot it might look something like this START putOnTable Bowl putOnTable Spoon putOnTable Napkin open Cereal pour Cereal open Milk pour Milk
223. cat s qfield setText s afield setText defrule ask ask question by id Given the identifier of a question ask it declare auto focus TRUE MAIN question ident id text text type type valid valid 212 CHAPTER 13 Adding a graphical interface not MAIN answer ident id MAIN ask id gt ask user text type valid engine waitForActivations defrule ask collect user input Check and optionally return an answer from the GUI declare auto focus TRUE MAIN question ident id text text type type not MAIN answer ident id Puser lt user input input ask lt MAIN ask id gt if is of type Pinput type valid then assert MAIN answer ident id text input retract ask user return else retract Pask user assert MAIN ask id Here s what happens when the new ask module is used 1 You assert a MAIN ask fact giving the identifier for a question 2 On the main thread the rule ask question by id is activated by the MAIN ask fact and the MAIN question fact with the given identifier This rule has the auto focus property so the ask module immediately gets the focus 3 ask question by id calls the function ask user and then uses waitFor Activations to pause until another rule in this module is activated 4 ask user sets up the GUI to display the proper question and clears the answer are
224. cations including using XML as a rule language and working with application servers Enter prise Java Beans and the J2EE environment The main text does not try to be an exhaustive guide to all of Jess instead it con centrates on those features relevant to the example applications The first two appendices provide some additional detail Appendix A includes a description of each of the functions built into the Jess language and appendix B presents the highlights of Jess s Java APIs The development methodology used in this book emphasizes testing Appen dix C presents a simple automated testing framework that can be used to test Jess applications The code for this framework is available from this book s web site xxviii ABOUT THIS BOOK Who should read this book Because this book can be used in several different ways it has several distinct pos sible audiences Part 1 is an introduction to rule based systems for any student of information technology practitioners and management alike The later parts of the book are aimed squarely at programmers I ve assumed an intermediate knowledge of the Java programming language throughout Occasionally I explain a Java concept but most of the time I just imagine that you understand The audience I thought of most as I wrote are intermediate Java programmers with little or no exposure to rule based systems who are interested in getting that exposure This book is also s
225. ce and you want to deduct expenses associated with that office The rules are fairly restrictive though you have to be careful or you will trigger an audit You usually shouldn t file this form unless you are self employed or your home is your primary workplace The accountant s expert knowledge is evident in a few of these answers particu larly the descriptions of Forms 2016EZ 4684 and 8829 Also evident however are some of the common problems with interview data The information is not very precise for example it is not true that everyone needs Form 1040 because there are two alternative forms After hearing these interview replies a few follow up questions suggest them selves immediately for instance you might want to confirm that the long version of Form 2016EZ is Form 2016 it is Otherwise you should be able to get the rest of the information from the forms themselves if you have questions about the forms you can arrange another interview 9 3 2 Reviewing the forms You can obtain copies of IRS forms from the IRS web site Studying the forms turns up a few potentially useful facts that the accountant didn t specify You can t file Form 1040EZ if you earned more than 400 in taxable interest People with more than a certain bank balance depending on current inter est rates might not be able to use this form You can only file Form 3903 if you changed work locations and if your new workplace is
226. ch loads Jess language code at runtime 7 All Java code which manipulates Jess entirely through its Java API The sample applications presented in the later parts of this book start at the beginning of this continuum the Tax Forms Advisor developed in part 3 consists entirely of Jess code and work their way toward the end the business rules sys tems are mostly written in Java Experiencing the development of each type of application will help you decide what route to take in future development based both on the requirements of the application and on the abilities of the program ming team that will write it 3 2 2 Jess performance Some people will tell you that Java is slow They re wrong Modern Java virtual machines are extremely powerful and sophisticated In many applications Java is as fast as comparable C or C code For Jess being written in Java is not a liability Jess is fast The algorithm used for pattern matching which we ll study in chapter 8 is very efficient and Jess can plow through large piles of rules and facts in little time Using Sun s HotSpot JVM on an 800 MHz Pentium III Jess can fire more than 80 000 rules per second it can perform almost 600 000 pat tern matching operations per second it can add more than 100 000 facts to working memory per second and a simple counting loop can do 400 000 itera tions per second Independent benchmarks have shown that Jess is significantly faster than many rule engin
227. chapter 18 Embedding Jess in Java applications In this chapter you ll a Learn about the jess Rete class a Create and manipulate facts from Java a Work with Jess exceptions Learn to reroute Jess s I O channels 306 Getting started with the Jess library 307 Every application you ve written so far has been based on the command line inter face tool jess Main The Tax Forms Advisor and PC Repair Assistant each con sisted entirely of a single Jess script whereas the HVAC Controller also included some Java classes Each of these three programs could be launched by starting jess Main and telling it the filename of a script from the command line It might look as though jess Main is Jess s central class In fact jess Main is just a command line wrapper around the Jess library With the library you can cre ate any number of individual Jess inference engines You can define rules for them add data to their working memories run them in separate threads and col lect generated results all from Java code without using jess Main The Jess library is obviously what you need to use to deploy rule based web applications Jess must be embedded inside some larger application a web server an application server or even a browser In this chapter you ll learn general tech niques for embedding Jess in Java software In the next chapter you ll put these techniques into practice and create a servlet based web application around
228. cit 120 explode 398 expressions 51 Boolean 52 evaluating 51 Extensible Markup Language 360 EXTERNAL_ADDRESS 309 422 external addressp 398 eXtreme programming 25 eye contact 154 F facade 307 420 FACT 422 fact base 21 75 fact id 78 398 Jess function 79 factorial 117 factory class 380 facts 75 295 399 as lists 75 creating 78 goal seeking 118 Jess function 77 177 ordered 76 81 86 query trigger 130 shadow 81 87 100 trigger 118 unordered 81 82 87 163 fact slot value 130 399 FALSE 43 390 fat client 288 290 fear of being replaced 155 feedback 184 fetch 309 399 findDeftemplate 311 findFactByld 314 findUserfunction 246 firePropertyChange 232 first 50 399 first best fit 404 FLOAT 422 float 399 floatp 399 flowcharts 152 188 191 197 216 256 drawing 189 focus 123 126 173 182 399 focus module 126 focus stack 127 128 182 for Java keyword 258 foreach 5 51 196 244 399 Forgy Charles 136 format 400 Freon 257 frontiers 151 FUNCALL 422 Funcall 316 Funcall NIL 245 Funcall TRUE 69 function calls evaluating 46 nested 46 results 46 functions deffunction 56 nesting 65 redefining 57 fuzzified 268 Fuzziness 265 fuzzy control system 264 fuzzy logic 256 264 fuzzy rules 268 fuzzy set 267 fuzzy set theory 266 fuzzy value 267 fuzzy variable 267 Fuzzy 265 Fuzzy Toolkit version 276 FuzzyJess 219 265 fuzzy match 268 FuzzySet 267 FuzzyValue 267 FuzzyVariable 267 G gam
229. class is shown in listing 14 5 Note how it uses State to convert between Boolean values and the symbolic constants open and closed The HeatPump class not shown is very similar Listing 14 5 A JavaBean to represent an automated vent package control public class Vent extends BeanSupport implements Runnable private Hardware m_hardware private int m_floor private boolean m_oldState public Vent Hardware hw int floor m_hardware hw m_floor floor new Thread this start public int getFloor return m_floor public String getState return State vent m_hardware getVentState m_floor public void setState String szState boolean state State vent szState m_hardware setVentState m_floor state m_pcs firePropertyChange state new Boolean m_oldState new Boolean state m_oldState state public void run while true boolean state m_hardware getVentState m_floor m_pcs firePropertyChange state new Boolean m_oldState new Boolean state m_oldState state 14 4 14 5 Summary 235 try Thread sleep 1000 catch InterruptedException ie return JavaBeans and serialization The notion of pickling is part of the JavaBeans concept The state of an application made out of JavaBeans is fully specified by the values of all the properties of those Beans together with information about the connections between the Beans A Java Beans f
230. code that won t tie you to a particular J2EE implementation Tomcat is open source you can download it from http jakarta apache org tomcat Although Pll explain important things as I go along I ve written this chapter with a fairly experienced Java programmer in mind Building a full web applica tion from scratch requires a range of knowledge and technical skills and there simply isn t enough room in this book to teach all of what you need to know Beside basic Java programming there s web browsing in general the Java Servlet API HTML and the HTTP protocol I have to assume you know at least a little about all these things If you ve been working with Java for a while this material should be easy for you to use If you re new to Java or the Web you may have a steeper learning curve In that case try not to get bogged down in the details read through it go off and hone your Java and web programming skills and then come back and read it again One more thing This is the largest application you ve built so far in this book and building and deploying it while not rocket science is more complicated than just typing java jess Main something clp Therefore I refer to some tools and software outside of Jess and Java The choices I ve made reflect my actual working environment the Linux operating system GNU make as a build tool and the Tomcat servlet container as an application server Your choices may be different You m
231. constructs and no other commands or constructs Jess Jess functions 403 parses the file and asserts each fact The return value is the return value of assert when asserting the last fact load facts looks for the named file the same way the batch command looks for scripts If the filename contains special characters it must be passed as a double quoted string load function lt lexeme gt Loads a class into Jess and adds it to the engine thus making the corresponding command available The argument must be the fully qualified name of a Java class that implements the jess Userfunction interface load package lt lexeme gt Loads a class into Jess and adds it to the engine thus making the corresponding package of commands available The argument must be the fully qualified name of a Java class that implements the jess Userpackage interface log lt number gt Returns the logarithm base of its only argument log10 lt number gt Returns the logarithm base 10 of its only argument long lt expression gt Interprets the expression as a Java long if possible and returns a LONG value Use strings or symbols for precise values that can t be expressed as an integer Longs in Jess are second class citizens in the sense that you can t directly do math on them You can assign them to variables pass them to function calls and convert them to strings or floating point numbers longp lt expression gt Returns TRUE if
232. ct Implied multislot __data FALSE In this example you can see the three special templates Jess uses internally __clear __fact and __not_or_test_CE as well as the initial fact template and your number template The third and final category of facts is perhaps the most interesting because these facts connect Jess s working memory to the real world outside Shadow facts A shadow fact is an unordered fact whose slots correspond to the properties of a Java Bean JavaBeans are a kind of normal Java object therefore shadow facts serve as a connection between the working memory and the Java application inside which Jess is running They re called shadow facts because they are like images or shad ows of JavaBeans outside of Jess The function defclass lets you create a special 88 6 5 1 6 5 2 CHAPTER 6 Representing facts in Jess deftemplate to be used with shadow facts and definstance lets you create an individual shadow fact Let s briefly look at what JavaBeans are and then see how they can be connected to Jess s working memory We ll discuss these Jess functions m defclass Creates a deftemplate from a JavaBean class m definstance Adds a JavaBean to Jess s working memory Jess and JavaBeans You can view Jess s working memory as sort of an electronic organizer for your rule based system A piece of data must be part of the working memory for it to be used in the premises of a Jess rule Orde
233. ct connected to one particular JavaBean instance into working memory Before you can insert an instance of DimmerSwitch into Jess s working memory you need a deftemplate to represent the class You use defclass to tell Jess to generate it Jess gt defclass dimmer DimmerSwitch DimmerSwitch Jess gt ppdeftemplate dimmer deftemplate MAIN dimmer extends MAIN __fact JAVA OBJECTS DimmerSwitch slot brightness default lt External Address jess SerializablePD gt slot class default lt External Address jess SerializablePD gt slot OBJECT type 2048 To run this code you first need to compile the DimmerSwitch class and make the class file available on your CLASSPATH The resulting template is a bit strange looking especially the default values It does have a slot called brightness as you d expect which arises from the brightness property of the JavaBean All shadow facts have the other slots in common The slot class comes from the method getClass that every Java object inherits from java lang Object and the slot OBJECT is added by Jess This special OBJECT slot always contains a reference to the JavaBean to which a given shadow fact The first argument to defclass dimmer here is used as the deftemplate name Some people like to use the name of the Java class itself my personal preference is to use a shorter name with a lowercase initial Putting a DimmerSwitch into working memory Now you can put a D
234. ct is on the fact list and if so to obtain a reference to it The argument is a Fact that is identical to the one you re looking for public Fact findFactByID int id Finds a fact given its numeric identifier This method is very slow don t use it unless you have to public Iterator listXXX Many methods return java util Iterators over collections of Jess constructs These include listActivations listDeffacts listDefglobals listDefinstances listDefrules listDeftemplates listFacts listFocusStack listFunctions listJessListeners and listModules Waiting for rules to fire Many multithreaded programs need to be able to pause indefinitely until Jess is ready to fire some rules These methods let you do that public Object getActivationSemaphore Returns the activation semaphore for this engine The monitor of the activation semaphore is signaled whenever an activation appears You can therefore call wait on this object and the wait call will return only when Jess is ready to fire a rule public void waitForActivations Doesn t return until there are rules on the agenda ready to fire B 9 jess RU RUstands for Rete Utilities This is a catch all class that contains some useful con stants Most of these constants enumerate the allowed return values for Value getType the different kinds of Value objects The most important ones are all static final int variables ATOM a symbol EXTERNAL_ADDRESS FACT FLOA
235. ctions and the equivalent Rete methods Table 18 1 Some simple methods in the jess Rete class Rete method Jess equivalent reset reset run run run int run number clear clear 18 2 Working with Fact objects in Java 311 For example in a code listing in the previous section you had engine executeCommand run Instead you can write this code as follows engine run The Rete class also includes direct Java equivalents of clear reset and many other important Jess functions In the following sections we ll look at what you can do with some more of these Rete methods Working with Fact objects in Java Rather than using executeCommand store and fetch to assemble and assert facts from Java you can construct jess Fact objects directly Ifa single fact is going to hold objects in several slots this method will be more convenient and it will always be more efficient Table 18 2 lists some of the Java methods that will help you work with Fact objects Table 18 2 Some jess Rete methods for working with facts Rete method Jess equivalent addDeftemplate Deftemplate deftemplate assertFact Fact assert fact findDeftemplate String None findFactByFact Fact None findFactBylId int fact id number retract Fact retract fact id The jess Fact class is a subclass of ValueVector you met ValueVector in sec tion 15 2 All the entries in the ValueV
236. d As a practical example you ll gather the knowledge you ll build into your first nontrivial rule based program The Tax Forms Advisor For the next three chapters you ll be developing a simple rule based application that recommends United States income tax forms The application asks the user a series of questions and based on the answers tells the user which paper Internal Revenue Service forms she will likely need You will populate the application with enough data to make it realistic although you won t try to make it exhaustive Your application might be used in an information kiosk at the post office You ll follow a realistic development process as you create this application start ing in this chapter by collecting the actual knowledge In chapters 10 and 11 you ll write the application using an iterative methodology including lots of testing The Tax Forms Advisor has a command line interface You ll concentrate on developing the rules themselves so the entire program will be written in the Jess rule language without using any Java reflection capabilities In the next part of this book we ll examine one way to add a graphical interface to applications like this one DISCLAIMER The system you re developing is intended only to provide guidelines about what tax forms a taxpayer might need to file It is not intended to give authoritative legal advice about tax filing Introduction to knowledge engineering Ever
237. d Jess in particular can be used with applications based on Enterprise JavaBeans Jess in the J2EE environment In this chapter you ll a Use Jess in EJBs a Write an RMI based rule service a Learn about the javax rules API 375 376 CHAPTER 21 Jess in the J2EE environment Most of the time when people talk about working in the J2EE environment they actually mean they are working with Enterprise JavaBeans EJBs Programming EJBs can be a real challenge but they offer powerful capabilities In this chapter we ll briefly review what EJBs are and how they are programmed and then we ll look at several strategies for incorporating rule engines in general and Jess in par ticular into EJB applications 21 1 A quick tour of EJB concepts 21 1 1 Enterprise JavaBeans is a component architecture for server side Java program ming The Beans referred to are the individual components They are the rela tively small relatively self contained pieces of code that implement your application EJBs are embedded in a container an application that provides essential services to the Beans These services include things like lifecycle manage ment creating and deleting beans network connectivity so client software can make requests on the beans database access persistence replication transac tions and more Kinds of EJBs There are three different kinds of EJBs entity beans session beans and message driven beans E
238. d destroy These methods are called at well defined times by the Java environment that hosts the applet invariably a web browser Writing applets that can work effectively inside a browser can be challenging because the user can browse away from the page containing an applet at any time only to return later An applet may or may not be destroyed during this time and on returning to the original page the user may find a new instance of the applet This lifecycle issue makes maintaining a consistent user session difficult To compound the problem web browsers vary widely in their level of Java sup port The major browsers include very old JVMs as standard equipment in fact some Internet Explorer users have no JVM installed JavaSoft offers a Java Plug in that provides a standard Java environment for browsers that use Netscape style plug ins and targeting this platform can be a good solution especially on an intranet But the Java Plug in is a large download and many Internet users won t have it and won t want to download it Dealing with the JVM support issue is the biggest obstacle to deploying substantial applets Jess supplies a simple example applet in the jess ConsoleApplet class Con soleApplet presents the same GUI as jess Console so it s useful only in limited situations applications that involve interview style interaction with a user Java Web Start The Java Web Start system lets a client click a link on a web page to do
239. d list of activated rules together with any other rules activated in previous cycles is called the conflict set 2 The conflict set is ordered to form the agenda the list of rules whose right hand sides will be executed or fired The process of ordering the agenda is called conflict resolution The conflict resolution strategy for a given rule engine will depend on many factors only some of which will be under the programmer s control 3 To complete the cycle the first rule on the agenda is fired possibly changing the working memory and the entire process is repeated This 2 3 2 2 3 3 Architecture of a rule based system 21 repetition implies a large amount of redundant work but many rule engines use sophisticated techniques to avoid most or all of the redun dancy In particular results from the pattern matcher and from the agenda s conflict resolver can be preserved across cycles so that only the essential new work needs to be done Many beginning rule programmers have difficulty with the idea that the rule engine will decide the order in which the rules will be fired but this is actually one of the great strengths of rule based programming The rule engine can more or less create a custom program for each situation that arises smoothly handling combinations of inputs the programmer might not have imagined The rule base Your rule engine will obviously need somewhere to store rules The rule base con tains all th
240. d by the store and fetch functions To send a Java object to Jess 1 Choose a unique identifier 2 In Java call Rete store passing the identifier and the object as arguments 3 In Jess call fetch using the same identifier to retrieve the object The color example would look like this import jess import java awt Color Rete engine new Rete Color pink new Color 255 200 200 engine store PINK pink Value v engine executeCommand assert color fetch PINK Fact f v factValue engine getGlobalContext The store method saves the Color object in a hash table using the string PINK as the key The fetch function retrieves the object from the same hash table Note that the store method is overloaded to accept either a jess Value or a generic java lang Object as its second argument The second version the one used here automatically wraps the Object in a jess Value of type RU EXTERNAL_ADDRESS before storing it Getting results If sending Java variables to Jess is the topic of the most frequently asked Jess ques tion the second most popular question surely concerns retrieving values from Jess back into Java Store fetch works in the other direction too It s a handy way of returning results from Jess to Java Here a value computed on the right hand side of a rule is retrieved from Java 310 18 1 3 CHAPTER 18 Embedding Jess in Java applications import jess Rete engine new Rete
241. d logic The concept of negation is a tricky thing It s tricky in Jess too Imagine that you want a rule to fire when no red cars are available Your first try might look something like this Jess gt defrule no red cars auto color red gt But this rule fires for each car that is not red If there are no cars at all it won t fire This result isn t the same as firing when there are no red cars Luckily Jess has the not conditional element Most patterns can be enclosed in a list with not as the head In this case the pattern is considered to match if a fact or set of facts that matches the pattern is not found For example this rule will fire if there are no cars at all or if there are only blue cars but not if there are any red ones 112 CHAPTER 7 Writing rules in Jess Jess gt defrule no red cars 2 not auto color red gt Because a not pattern matches the absence of a fact it cannot define any variables that are used in subsequent patterns on the LHS You can introduce variables in a not pattern as long as they are used only within that pattern Jess gt defrule no odd numbers not number n amp oddp n gt printout t There are no odd numbers crlf Similarly a not pattern can t have a pattern binding again because it doesn t match an actual fact there would be no fact to bind to the variable Now here comes the tricky part I alluded to earlier You already know that pat tern
242. data processing rules into another and your reporting rules into a third Then changing the focus from input to processing to output represents a natural pro gression through well defined phases of your application s execution The auto focus declaration When a rule that declares the auto focus property is activated its module auto matically gets the focus as illustrated in listing 7 6 In this example the rule crash fires even though its module PROBLEMS didn t have the focus and the agenda of the previous focus module DRIVING was not empty Modules with auto focus rules make great background tasks in conjunction with using return from a rule as described next Jess gt defmodule PROBLEMS TRUE Jess gt defrule crash declare auto focus TRUE DRIVING me location DRIVING other car location gt printout t Crash crlf halt TRUE Jess gt defrule DRIVING travel me lt me location gt 128 7 7 CHAPTER 7 Writing rules in Jess printout t retract me assert me location 1 TRUE Jess gt assert me 1 lt Fact 1 gt Jess gt assert other car 4 lt Fact 2 gt Jess gt focus DRIVING MAIN Jess gt run Crash 4 Returning from a rule s RHS If the function return is called from a rule s RHS the execution of that rule s RHS is immediately terminated Furthermore the current focus module is popped from the focus stack This su
243. de The important thing to realize here is that any Java code can call myMethod without knowing that it s a native method From the caller s perspective a native method looks like any other Java method A subclass can override a normal parent method using a native method and an implementation of an interface can pro vide a native method to satisfy the interface You can use this last fact to your advantage You ll define a Java interface to represent the functions in the C library Then you can write multiple implementa tions of the interface one that uses native methods to access the real hardware and one that implements a simulation of the hardware You can then use the sim ulator to develop and test the software When the software is ready you could swap in the hardware implementation and try it out on the real system For the purposes of this book you ll only write the simulator The interface has one method for each of the C library functions so it looks like this package control public interface Hardware public final int OFF 0 HEATING 1 COOLING 2 int getHeatPumpState int heatPump void setHeatPumpState int heatPump int state int getNumberOfHeat Pumps double getTemperature int sensor boolean getVentState int vent void setVentState int vent boolean state Note that you ll put all the Java code you write for this application into a package named control The methods in this interface look
244. ded an exhaustive list of my embarrassingly many typos in the first draft of the manuscript Thanks Bob for everything Next I must mention the denizens of the Jess mailing list a friendly community of smart and generous people who have come together over the years that Jess has existed The following people have helped find bugs helped develop new features xxiii Xxiv ACKNOWLEDGMENTS or contributed their own projects to the Jess community Abel Martinez Al Davis Alan Moore Alex Jacobson Alex Karasulu Andreas Rasmusson Andrew Marshall Ashraf Afifi Benjamin Good Blaine Bell Bob Orchard Bob Trelease Bruce Dou glas Chad Loder Charles May Cheruku Srini Dan Larner Dave Barnett Dave Carlson Dave Kirby David Bruce David Li David Young Drew van Duren Duane Steward Ed Katz Emmanuel Pierre Eric Eslinger Fang Liu George Rudolph Glen Tarbox Glenn Williams Henrik Eriksson Ian de Beer J P van Werkhoven Jacek Gwizdka Jack Fitch Jack Kerkhof James Gallogly James Owen Jason Smith Javier Torres John Callahan John Collins Joszef Toth Juraj Frivolt Karl Mueller Ken Bertapelle Kenny Macleod Lakshmi Vempati Lars Rasmusson Laurence Leff Mariusz Nowostawski Matt Bishop Matthew Johnson Michael Coen Michael Friedrich Michael Futtersack Michal Fadljevic Michelle Dunn Mikael Rundqvist Mike Finnegan Mike Isenberg Mike Lucero Miroslav Madecki Nancy Flaherty Ning Zhong Norman Ghyra Oliver Hoffman Osv
245. dent sound text yes answer ident seek text yes answer ident boot begins text yes gt recommend action halt 12 5 4 RAM problems consult a software expert The RAM flowchart shown in figure 12 5 is simple it includes only two more rules One handles the case where the RAM memory modules are loose in their sockets and the other handles the case where they are not These two rules are slightly different than all the others in that they will be asked aftera recommended action has already been taken the Check RAM recommendation from rule check ram defrule MAIN loose ram declare auto focus TRUE check loose ram ident loose ram answer gt text yes recommend action halt defrule MAIN faulty ram declare auto focus TRUE remove and reseat memory modules 200 CHAPTER 12 Writing the PC Repair Assistant check loose ram answer ident loose ram text no gt recommend action replace memory modules one by one and retest halt START Is Remove RAM YES and reseat loose memory modules NO LA Replace memory modules one by one Figure 12 5 The diagnostic flowchart for computers that beep fewer than END three times and then don t boot 12 5 5 Questioning authority Now that you ve written all the rules you need the questions Listing 12 6 shows them collected in a deffacts Listing 12 6 Que
246. discussed in section 6 4 It is important to recognize that we re using the word fact in a specific techni cal sense and the meaning differs slightly from the colloquial English usage In rule based systems terminology fact is another word for working memory element A fact is therefore the smallest unit of information that can be separately added to or removed from the working memory of a rule based system Jess facts aren t gen erally atomic rather they have some structure to them as you ll see in the follow ing sections Manipulating the working memory Intuitively a collection is a group of items you can add to or remove from Depending on your tastes as a collector you can acquire new baseball cards or discard a chipped ceramic unicorn Working memory is a collection of facts and if you were to write a rule based program about trading cards each card in your collection might be represented by a single fact Jess offers a set of functions to let a program perform the basic collection oper ations add remove modify duplicate on the working memory You can also use these functions interactively while you re programming along with others that let you examine initialize and clear the working memory This section describes the following functions and constructs as a Jess programmer you ll use all of these frequently m assert Adds facts to working memory m clear Clears all of Jess m deffacts Defines the initial
247. doesn t fire because the DRIVING module doesn t have the focus You can move the focus to another module using the focus function which returns the name of the previous focus module Jess gt focus DRIVING MAIN Jess gt run Ready to go 1 Partitioning the rule base with defmodule 127 Note that you can call focus from the RHS of a rule to change the focus while the engine is running The focus can move many times during a single run of a program Jess maintains a focus stack containing an arbitrary number of modules The focus command pushes the new focus module onto the top of this stack the focus module is by definition the module on top of the stack When there are no more activated rules in the focus module it is popped from the stack and the next mod ule underneath becomes the focus module The module MAIN is always at least on the bottom of the stack it can also be explicitly pushed onto the focus stack You can manipulate the focus stack directly with the functions pop focus clear focus stack list focus stack and get focus stack pop focus removes the focus module from the focus stack so that the next module on the stack becomes active clear focus stack removes all the modules from the focus stack The other functions let you examine the contents of the focus stack Rule bases are commonly divided into modules along functional lines For example you might put all your input gathering rules into one module your
248. dress gt lt jess code gt Executes some Jess code in the context of the given Rete object the external address This is a nice way to send messages between multiple Rete engines in one process Note that the current variable context is used to evaluate the code so for instance all defglobal values will be from the calling engine not the target Jess gt bind engine new jess Rete lt External Address jess Rete gt Jess gt call on engine Pengine 2 2 4 clear Clears the rule engine Deletes all rules deffacts defglobals templates facts activations and so forth User functions written in Java are not deleted clear focus stack Removes all modules from the focus stack clear storage Clears the hash table used by store and fetch close lt lexeme gt Interprets each argument as an I O router name Closes any I O routers associ ated with the given names by calling close on the underlying stream and then removes the routers Any subsequent attempt to use a closed router will report bad router See open 396 APPENDIX A Jess functions complement lt list gt lt list gt Returns a new list consisting of all elements of the second list not appearing in the first list context Returns the execution context a jess Context object in which it is called This provides a way for deffunctions to get a handle to this useful class count query results lt lexeme gt lt expression gt Runs th
249. ds to embedded hardware Before you begin to code your rules you try to develop a picture of what your system will need to realize these connections Depending on your development environment your rules may already have a built in ability to connect to all the data sources and sinks they ll need to reach directly from the rule language In other situations you may need to write interface code in another language If you do I hope you ll use test first programming to develop it We ll look at interface building many times throughout this book 2 4 5 Writing the rules Once the data structures are defined the interfaces are specified and the tests are in place it s time to begin writing the rules As in all programming this pro cess involves a significant amount of art there are always multiple ways to accomplish a task The good news is that because each rule can be independent of the others rule based programs can be developed iteratively code a little test a little and then code some more The bad news is that it s relatively easy to write unstructured rule based programs which can become hard to understand 4 K Beck Extreme Programming Explained Embrace Change Reading Mass Addison Wesley 2000 26 2 4 6 2 5 CHAPTER 2 What are rule based systems You can give structure to your rule based programs by thinking in terms of phases or modules groups of rules that are the only ones relevant at specific phases
250. dule names on the focus stack as a list The top module on the stack is the first entry in the list get member lt object gt lt lexeme gt lt symbol gt Retrieves the value of a Java member variable The first argument is a Java object or the name of a class for a static member and the second argument is the name of the variable The return value is converted to a suitable Jess value exactly as for call get multithreaded io Returns TRUE if Jess is currently using a separate thread to flush I O streams Turning this on can lead to a modest performance enhancement but may result in loss of output on program termination Jess functions 401 get reset globals Indicates the current setting of global variable reset behavior See set reset globals for an explanation of this property get salience evaluation Indicates the current setting of salience evaluation behavior See set salience evaluation for an explanation of this property halt Halts the rule execution cycle Has no effect unless the rule engine is running if lt boolean gt then lt expression gt else lt expression gt Allows conditional execution of a group of actions The Boolean expression is eval uated If it does not evaluate to FALSE the first list of expressions is evaluated and the return value is whatever is returned by the last expression of that list If the Boolean expression evaluates to FALSE and the optional second list of expres
251. e foreach Pitem valid printout t Pitem printout t bind answer read return answer deffunction is of type answer type valid Check that the answer has the right form if eq type multi then foreach Pitem valid if eq sym cat answer sym cat item then return TRUE return FALSE if eq type number then return is a number answer plain text return gt str length Panswer 0 deffunction is a number value Return TRUE if value is a number try integer value return TRUE catch return FALSE E This version of the is of type function has to do more work than the version in section 10 6 because it needs to check three different kinds of questions and it can t assume the answers will arrive as Jess numbers and symbols every answer might be in the form of a string and need to be parsed further When checking multi answers is of type therefore uses the sym cat function to convert the answer to a symbol before comparing it to each possible result When checking numeric answers is of type calls another new function is a number to do the work The new is a number utility function uses Jess s built in integer function which either returns an integer version of its argument or throws an exception if the argument can t be interpreted as a number The is a number function uses 12 5 12 5 1 The rest of the rules 197 the Jess try fu
252. e abound in business environments A human resources application may need to flag personnel with a suspicious pattern of insurance claims A requirement for a financial application might be to recommend buying securities that look promis ing Manufacturing software could raise an alarm and shut down an assembly line if quality assurance results indicate there may be a production problem 1 2 1 2 1 Some real world examples 9 What do suspicious promising and may be a problem mean exactly They proba bly can t be expressed as equations it s possible they can t be given a precise def inition at all If they each can be described as a set of constraints and guidelines however then a rule based program can implement them easily In the rest of this chapter we ll look at some common applications of rule based programming Some real world examples You are probably affected by rule based software every day whether you realize it or not Many commonplace activities in a modern office are controlled by rules Let s look at a few examples Mail filtering I can t keep up with all the email I get is a common complaint in this Net con nected world Huge numbers of email messages are sent delivered and read every day Rules to the rescue There are many technological solutions to the prob lem of sorting email and almost without exception these solutions are simple rule based systems The venerable program sendma
253. e a function at any time and any code that called the old function will automatically call the new one To redefine a function you simply define a new function by the same name and the old one is replaced You can even redefine any of the built in 58 4 5 CHAPTER 4 Getting started with the Jess language functions this way although doing so is not recommended If you want to change the behavior of a built in function you can do so using defadvice described in the next section Fine tuning a function s behavior If you re a Java programmer it probably bothers you that functions like nth num ber list items starting at one rather than zero In some languages you d be stuck But Jess makes it easy to fine tune the behavior of any function so if you d like nth to use zero based indices you can have it your way by using advice Advice is code that you add to a function to change its behavior The function defadvice lets you easily wrap advice around any Jess function such that it executes before and thus can alter the argument list seen by the real function or short circuit it completely by returning a value of its own or after the real function and thus can see the return value of the real function and possibly alter it This is a great way for Jess add on authors to extend Jess without needing to change any internal code It was named after a similar feature in Emacs Lisp To make nths accept a zero based index all y
254. e any INTEGER or FLOAT public Value int value int type Constructs a numeric value The type can be INTEGER or FLOAT Uses the class LongValue to construct a Value of type LONG 424 APPENDIX B Abridged Java API for Jess public Value java lang Object o Constructs a value containing a Java object type EXTERNAL_ADDRESS public Value java lang String s int type Constructs a lexeme value either STRING or ATOM public Value Value v Copies a value public Value ValueVector f int type Constructs a value of type LIST public int type Returns the type of this variable one of the constants in RU public Value resolveValue Context c Given an evaluation context returns the true value of this Value For variables this means the value of the variable For Value objects containing function call objects the function is called and the return value provided public String atomValue Context c Returns the contents of this value as a symbol public Object externalAddress Value Context c Returns the contents of this value as a Java object public Fact factValue Context c Returns the contents of this value as a fact public double floatValue Context c Returns the contents of this value as a number public Funcall funcallValue Context c Returns the contents of this value as a function call public int intValue Context c Returns the contents of this value as an int public ValueVector listValue Context c
255. e are printed The symbol indicates all modules With no argument only the facts from the current module are listed fact slot value lt fact id gt lt symbol gt Returns the value in the named slot of the fact with the given fact id fetch lt lexeme gt Retrieves and returns any value previously stored by the store function under the given name or nil if there is none Analogous to the fetch member function of the Rete class Note that the storage used by store and fetch is local to each individual Rete object first lt list gt Returns the first field of a list as a new single element list like car in traditional LISP float lt number gt Converts its only argument to a Jess floating point number floatp lt expression gt Returns TRUE for arguments of type RU FLOAT otherwise returns FALSE focus lt symbol gt Changes the focus module The next time the engine runs the first rule to fire will be from the first module named as an argument if any rules are activated in this module The previously active module is pushed down on the focus stack If more than one module is listed they are pushed onto the focus stack in order from right to left If the engine is running at the time of the call the focus changes immedi ately and the next rule to fire will be from the first named module foreach lt variable gt lt list gt lt expression gt Sets the named variable to each of the values in the list i
256. e based systems are commonly used in web based applications A Java rule engine can run on a client machine as an applet or more commonly on the server in a J2EE application server Rules can be used for order configuration inventory management human resources manufacturing resource planning games and more In part 6 of this book you ll develop a Recommendations Agent for an online store The Recommendations Agent will look at the items a customer is buying and has bought in the past and make recommendations for additional pur chases In this chapter we ll look at an overview of web architectures and then develop the data structures and rules for the agent In chapter 18 you ll learn how to embed Jess as a library in a larger application and we ll study techniques for interacting with Jess through its Java API In chapter 19 you ll package the Recommendations Agent as a web application deployed in a standard J2EE con tainer The application will be built from servlets and JavaServer Pages see sec tion 17 1 2 for definitions You ll build a complete e commerce web site in miniature to serve as an example platform for the Recommendations Agent although you ll leave out the bits that are entirely unrelated to the agent such as credit card validation 17 1 Java architectures for the Web 17 1 1 There are many different ways to deploy applications on the Web In this part of this book I ll use the phrase the Web to r
257. e customer is buying or has bought in the past a VCR or DVD player and when the customer is buying or has bought in the past a videotape or DVD You can handle the first case recommending a tape or disk to go with a new player by adding the appropriate category to the requires multislot of each player in the catalog The other three cases require adding spe cific rules First let s look at the rule that recommends new media to anyone who has ever bought a player in the past defrule recommend media for player If customer has bought a VCR or DVD player in the past recommend some media for it product part number media category c amp dvd disk videotape product name name part number player category if eq c dvd disk then dvd else vcr The first pattern matches any DVD or videotape and the second pattern matches all the players that can play it These two patterns include some fancy matching The first pattern uses the or connective constraint to match either dvd disk or videotape in the category slot The second pattern uses the return value constraint and a conditional expression to select a value for the category slot that goes with the media category vcr for videotape and dvd for dvd disk We ve deliberately put these two patterns at the beginning of this rule because they match only things from the catalog and therefore will never change while the program is running It s generally good to put th
258. e entry at position i to val 426 APPENDIX B Abridged Java API for Jess public ValueVector setLength int i Sets the size of the list Value objects at this index and above are no longer part of the list public int size Returns the number of Value objects in this ValueVector An automated testing framework 428 APPENDIX C An automated testing framework When you re doing iterative development automated testing gives you confidence that each addition or modification does not introduce errors into a previously work ing program You make a change you run your tests and you find out immediately whether your changes broke the existing code Having a good suite of automated tests that you can run quickly and easily helps you write better code more quickly This is as true for rule based programming as it is for any other kind Whether I m working on extensions to Jess or writing rule based programs I find that auto mated testing helps keep the quality of the software I write as high as it can be In this appendix I ll describe a framework I use to write tests for Jess language code The actual code for the framework is included with the sample code for this book Note that for Java application programming server side Java programming and programming in many other languages a number of excellent free testing frameworks are available JUnit http wwwjunit org index htm was the first of a large family of these frameworks
259. e media rules There s one more marketing rule you haven t written yet the rule recommend same type of media which tries to get a past media customer to buy more media on every return visit It s really just a combination of recommend more media and recommend media for player so I ll present it here without any narration defrule recommend same type of media If customer has bought a disk or tape in the past recommend a random other item of the same category There are two recordings of the same type pl lt product part number part1 category c amp dvd disk videotape name name p2 lt product part number part2 category c test neq p1 p2 300 17 6 CHAPTER 17 Jess on the Web This customer has bought one of them in a past order line item customer id id order number orderl part number part1 order customer id id order number currentOrder amp gt currentOrder orderl1 not order customer id id order number order3 amp gt order3 currentOrder 7 But not the other not line item customer id id part number part2 and we haven t recommended any media of this type yet not recommend order number currentOrder type sym cat discretionary c gt Recommend the other one assert recommend order number currentOrder because name part number part2 type sym cat discretionary c Refining the recomme
260. e of controlling the heating ventilation and air conditioning systems in a large office building For the first time you wrote an application that integrates code written in Java with code written in the Jess language You wrote a simulator to model the hardware your application is intended to control The simulator runs in its own thread You wrote JavaBeans and used them as working memory elements and each Bean also includes its own thread for poll ing the state of the hardware device it stands in for You saw how Jess works well in this multithreaded environment You learned how to extend the Jess language by writing Userfunctions and Userpackages Many people have written Jess extensions and then made them publicly available Examples include extensions for working with databases with XML and with fuzzy logic See the Jess web site for details on obtaining these and other Jess extensions You developed several different sets of control rules one conventional set and one using fuzzy rules The fuzzy rules although more complex controlled the temperature more tightly Although much of the HVAC Controller system application was written in Java the main application code is written as a Jess script and runs with jess Main In 284 CHAPTER 16 Writing the rules the next part of this book you ll integrate rules into a web based application where Jess will run as part of a Java servlet You won t use the jess Main class and so you ll n
261. e patterns that match fixed unchanging facts at the top of a rule so they won t need to be evaluated more than once In general ordering patterns is a balancing act with at least three important interacting factors Clarity You wrote the patterns in a rule in a specific order because that s the way they made sense Performance You certainly want to avoid doing repetitive work ordering patterns to put the static ones at the top of the rule as you ve done here helps you achieve this goal Memory use If you put patterns that match fewer facts toward the begin ning of a rule you ll reduce the number of partial matches and limit mem ory consumption 298 17 5 4 CHAPTER 17 Jess on the Web Often these three considerations conflict and you have to make a judgment call In this chapter I ve ordered the patterns placing the most emphasis on readabil ity In general this is a good idea at first you can always go back and reorder the patterns to increase performance and reduce memory usage later if it turns out to be necessary The following three patterns identify a customer who has bought a player in the past and is placing a new order The line item pattern identifies a purchase of a player matched in the first part of the rule The next two patterns identify the cus tomer whose past order contained that line item and who is placing a new order line item customer id id order number order1 part nu
262. e query whose name is given by the first argument and returns a count of the matches See the documentation for defquery for more details Also see run query for caveats concerning calling count query results from a rule create lt expression gt Returns a new list containing all the given arguments in order For each argu ment that itself is a list the individual elements of the list are added to the new list this function will not create nested lists which are not meaningful in the Jess language Note that lists must be created explicitly using this function or others that return them ordinary lists cannot be directly parsed from Jess input defadvice before after lt lexeme gt lt list gt lt expression gt Lets you supply extra code to run before or after the function s named by the second argument The list of expressions is taken as a block of code to execute If before is specified the code executes before the named function s the variable argv holds the entire function call vector function name and parameters on entry to and exit from the code block If after is specified the function is called before the code block is entered When the block is entered the variable retval refers to the original function s return value Whether before or after is specified if the code block explicitly calls return with a value the returned value appears to the caller to be the return value of the original function For
263. e rules the system knows They may simply be stored as strings of text but most often a rule compiler processes them into some form that the inference engine can work with more efficiently For an email filter the rule compiler might produce tables of patterns to search for and folders to file messages in Jess s rule compiler builds a complex indexed data structure called a Rete network A Rete network is a data structure that makes rule processing fast Chapter 8 describes how Jess s rule compiler works In addition the rule compiler may add to or rearrange the premises or con clusions of a rule either to make it more efficient or to clarify its meaning for automatic execution Depending on the particular rule engine these changes may be invisible to the programmer Some rule engines allow or require you to store the rule base in an external relational database and others have an integrated rule base Storing rules in a relational database allows you to select rules to be included in a system based on criteria like date time and user access rights The working memory You also need to store the data your rule engine will operate on In a typical rule engine the working memory sometimes called the fact base contains all the pieces of information the rule based system is working with The working memory can hold both the premises and the conclusions of the rules Typically the rule engine maintains one or more indexes similar to t
264. e they good for Where did they come from Are they right for you What should you do if you want to build one These are the questions we ll begin to address in part 1 You ll learn what rule based sys tems are about their history and about their many uses We ll also look at how to decide when a rule based solution is appropriate for your application Finally you ll learn about how rule based systems are implemented and some strategies for developing them Rules to the rescue In this chapter you ll Be introduced to the Jess programming language m Analyze a rule based program See familiar examples of rule based systems 11 CHAPTER 1 Rules to the rescue Rule based software is in regular use in practically every business school and home In this chapter we ll look at some examples of how rules are used to solve common problems Because most programmers learn best by doing you ll start by writing a rule based program of your own Math class melee The answer please The stern voice startles you You were dozing in Mrs Rosencrantz s high school math class again You realize at once that she s been talking to you Well You look at the blackboard It s one of those word puzzles the logic kind Mrs Rosencrantz is waiting for you to solve it You quickly scan what she s scrawled on the board with her crone s hand m A foursome of golfers is standing at a tee in a line from left to
265. e to TekMart com lt TITLE gt lt HEAD gt 336 CHAPTER 19 Deploying web based applications lt BODY gt lt H1 gt Welcome to TekMart com lt H1 gt Welcome to the TekMart com online store lt P gt To get started please enter your customer id lt FORM action Order catalog method POST gt lt INPUT name customerId gt lt INPUT type submit value Log in gt lt FORM gt lt P gt lt BODY gt lt HTML gt There s not much to see here assuming you re familiar with HTML forms The customer sees an input field and a button labeled Log In and can enter his login name and click the button Figure 19 2 shows what the form looks like The action Order catalog line gives the URL to which the result of submitting this form will be sent you ll name this application Order and catalog is the first serv let The host name port number and other details are assumed by the web browser that submits the form to be the same as for the JSP Welcome to TekMart com Welcome to the TekMart com online store To get started please enter your customer id Figure 19 2 The TekMart login screen as implemented in the file index htm1 The customer ID will be used by the serviets in the Recommendations Agent The Catalog servlet 337 19 8 The Catalog servlet The tasks the Catalog servlet has to do were listed in section 19 5 PII repeat them here slightly refined 1 Get the login name as entered
266. ean id queryResult class java util Iterator scope request gt lt HEAD gt lt TITLE gt Some recommendations for you from Tekmart com lt TITLE gt lt HEAD gt lt BODY gt lt H1 gt Your Recommendations lt H1 gt You may also wish to purchase the following items lt FORM action Order purchase method POST gt lt TABLE border 1 gt lt TR gt lt TH gt Name lt TH gt lt TH gt Catalog lt TH gt lt TH gt Because you bought lt TH gt lt TH gt Price lt TH gt lt TH gt Purchase lt TH gt lt TR gt lt while queryResult hasNext Token token Token queryResult next Fact factl token fact 1 Fact fact2 token fact 2 String partNum fact2 getSlotValue part number stringValue null gt lt TR gt lt TD gt lt fact2 getSlotValue name stringValue null gt lt TD gt lt TD gt lt partNum gt lt TD gt lt ValueVector vv fact1 getSlotValue because listValue null for int i 0 i lt vv size i gt lt vv get i stringValue null gt lt if i vv size 1 gt lt gt lt TD gt lt TD gt lt fact2 getSlotValue price floatValue null gt lt TD gt lt TD gt lt INPUT type checkbox name items value lt partNum gt gt lt TD gt lt TR gt 348 CHAPTER 19 Deploying web based applications lt gt lt TABLE gt lt INPUT type submit value Purchase gt lt FORM gt
267. earning about the Jess software itself in chapter 3 and proceed from there 5 See for example http www dfki uni kl dc ruleml Part 2 Jess A rule based programming environment L going to develop rule based systems first you ll need to pick a rule engine For the rest of this book you ll be working with the Jess rule engine that you first met in chapter 1 In chapter 3 you ll learn about Jess s origins how to get a copy and how to write the Jess version of the famous Hello World program In the next few chapters you ll learn a lot about writing programs in Jess s rule lan guage Chapter 4 is a general introduction to the Jess language Chapter 5 teaches you how to work with Java objects from Jess programs The next two chapters talk about Jess s working memory and how to write rules respectively Finally chapter 8 explains some of the nuts and bolts that make Jess work All together part 2 of this book is a comprehensive tutorial on both the Jess language and the Jess rule engine it will get you ready to start developing rule based applications in part 3 Introducing Jess In this chapter you ll m Learn how to obtain and install your copy of Jess m Learn how to run Jess programs m See how Jess applications are structured 31 32 3 1 CHAPTER 3 Introducing Jess This is a hands on book It walks you through the development of several large software systems that use rule based technology
268. ears in the list of defined functions or FALSE otherwise The two examples confirm that it works Summary The Jess language has a simple and regular syntax It s rather different from Java s syntax but it s easy to learn It is a full featured programming language in which you can write real programs Given that it is an interpreted language it is surpris ingly fast The Jess language has only a few built in data types including INTEGER FLOAT SYMBOL STRING and LONG There are several simple control structures some of which let you transform data into executable code All of the control structures are actually functions Jess comes with almost 200 built in functions and you can define your own functions using the deffunction construct You can modify the behavior of the built in functions using defadvice On top of all that you can also call any Java function you want from Jess How to do so is the topic of the next chapter Scripting Java with Jess In this chapter you ll Create Java objects Call Java methods Access Java member data a Gi Gi m all from the Jess language 61 62 5 1 CHAPTER 5 Scripting Java with Jess Arguably the most powerful features of Jess are those that allow it to be easily inte grated with Java From Java code you can access all parts of the Jess library so that it s very easy to embed Jess in any Java application servlet applet or other system Likewise from the
269. ecause Jess has to examine each Fact in working memory until it finds the right one Therefore there s no cheap way to remove a fact from working memory if you don t already have a reference to the Fact object Either Jess has to do a slow lookup or you have to do the work of constructing a duplicate Fact object If you assert Facts from Java and intend to retract them later in your application be sure to keep references to the Fact objects you create Working with JavaBeans Adding JavaBeans to Jess s working memory from Java is simple and it works pretty much the same way as it does from the Jess language Table 18 3 shows the correspondence between methods in the Rete class and the Jess functions you ve already learned about 18 4 Calling Jess functions from Java 315 Table 18 3 jess Rete methods for working with JavaBeans Rete method Jess equivalent defclass tag class name parent defclass tag class name parent definstance tag object boolean definstance tag object static dynamic undefinstance object undefinstance object You use the Rete defclass method to register a class with Jess and then you use Rete definstance to add individual instances to working memory You can remove the objects using Rete undefinstance As an example you can add a java awt Button to Jess s working memory like this Rete engine engine defclass button java awt Button null Button b new B
270. ecial kind of rule with no RHS Jess controls when regular rules fire but queries are used to search the working memory under direct program Searching working memory with defquery 129 control A rule is activated once for each matching set of facts whereas a query gives you a java util Iterator of all the matches An example should make this clear Suppose you have defined the query find affordable gifts Jess gt deftemplate gift slot name slot price TRUE Jess gt defquery find affordable gifts Finds all gifts in a given price range declare variables lower upper gift price p amp and gt p Plower lt p upper TRUE The pattern here matches all the gifts whose price slot holds a number between lower and upper Now you define some facts including some that match the criterion and some that don t Jess gt deffacts catalog gift name red scarf price 20 gift name leather gloves price 35 gift name angora sweater price 250 gift name mohair sweater price 99 gift name keychain price 5 gift name socks price 6 gift name leather briefcase price 300 TRUE You can invoke the query to find the perfect gift using concrete upper and lower price limits Jess gt reset TRUE Jess gt bind it run query find affordable gifts 20 100 lt External Address java util AbstractListSItr gt Jess gt while it hasNext bind token call it next bind fact
271. ector correspond to slots of the Fact the data for the first slot is the item at index 0 The head or name of the fact is stored in a separate variable and is accessible via the getName method Every jess Fact has an associated jess Deftemplate object that describes the slots the fact can have All Facts with a given head should share the same Deftem plate When you re programming in Java it s up to you to enforce this require ment If the Deftemplate you need has already been defined you can use the findDeftemplate method in Rete to get a reference to it otherwise you ll have 312 CHAPTER 18 Embedding Jess in Java applications to construct it yourself Once you ve built a Deftemplate you can reuse it for all the other Facts of the same type that you create Listing 18 1 shows how to create a Deftemplate with two slots and then assert several facts that use it Deftemplate s first two constructor arguments are the name of the template and the documentation string for the template The argu ments to addSlot are a name for the slot a default value for the slot and the name of the data type of the slot respectively Jess doesn t do anything with the data type argument it s reserved for future use import jess public class CreateFacts public static void main String unused throws JessException Rete engine new Rete Deftemplate d new Deftemplate person A person engine d addSlot name Funca
272. ectronic media You can often find hundreds or even thousands of references by typing a few key words describing your domain into an Internet search engine There are online encyclopedias of every description guides to technical fields troves of engineer ing data and countless other valuable resources Although the Internet is full of information it is important to realize that not all of it is correct or unbiased In particular many search engines either accept payment for highly placed listings or use a ranking system that is easily fooled into placing a particular page at the top of your search results Before using a general search engine learn a little about how it is implemented and operated Select one that to the extent possible ranks results only on their relevance to your search topic You should also scrutinize individual web pages check for the source of the information and try to verify it against another reference Sometimes published electronic reference works on CD ROM are useful although they are often simply expensive alternatives to or worse a repackaging of material already available on the Web Again you can often find and use these references in libraries Collecting knowledge about tax forms The domain for the example program is distributing income tax forms The project sponsors might describe it like this The system should ask the user a series of questions and then recom mend a list of income t
273. ects are independent 382 21 2 3 21 2 4 21 2 5 CHAPTER 21 Jess in the J2EE environment Implementing a main method When you write an RMI server you need to supply a main method that creates any objects that should exist when the server starts up The main method then regis ters those objects with a naming service so clients can find them The naming ser vice can either be the standalone program rmiregistry that comes with the J2SE or an enterprise directory service like LDAP Here you ll use rmiregistry accessed via the java rmi Naming class Your main will create a single JessFacto ryImp1 object and register it under the name Jess You can put this method in the JessFactoryImp1 class public static void main String argv throws Exception JessFactoryiImpl impl new JessFactoryImpl Naming rebind Jess impl System out println Jess server running Here you ignored the fact that most RMI servers install a special RMI security manager a full implementation would certainly do so You won t use one here because doing so would complicate the deployment step by requiring you to write a security policy file Generating the stubs The rmic stub compiler operates on the compiled implementation classes not as you might expect on the remote interfaces Once you have compiled JessImp1 and JessFactoryImpl you can run rmic like this rmic JessImpl JessFactoryImp1l This command creates JessImpl_Stub class Jess
274. ed automatically In the simple example in listing 7 3 the water flowing fact is again logically dependent on the faucet open fact so when the latter is retracted the former is removed too Jess gt clear TRUE Jess gt defrule water flows while faucet is open logical faucet open gt assert water flowing TRUE Jess gt assert faucet open lt Fact 0 gt oo Eum Rule water flows while faucet is open fires Jess gt facts f 0 MAIN faucet open f 1 MAIN water flowing 116 7 4 CHAPTER 7 Writing rules in Jess For a total of 2 facts Jess gt watch facts TRUE Jess gt retract 0 lt f 0 MAIN faucet open lt f 1 MAIN water flowing Jess retracts fact water TRUE flowing automatically Jess gt facts For a total of 0 facts If fact 1 is logically dependent on fact 2 you can also say that fact 1 receives logi cal support from fact 2 A fact may receive logical support from multiple sources it may be asserted multiple times with a different set of logical supports each time Such a fact isn t automatically retracted unless each of its logical sup ports is removed If a fact is asserted without explicit logical support it is said to be unconditionally supported If an unconditionally supported fact also receives explicit logical support removing that support will not cause the fact to be retracted You can find out what logical support a fact is rece
275. ed it When I typed yes instead the function returned yes Fetching the question text The question template has a slot to hold the text of a question and another slot to hold a unique identifier Similarly the answer template associates that same identifier with an answer You d like to call something from the right hand side RHS of a rule in the interview module using just the identifier and have that something look up the question text ask the question and assert an answer fact There are two ways to fetch something in working memory using a defquery or using a defrule Of the two rules are cheaper computationally because invok ing a query always involves clearing part of the Rete network and asserting one or Building the infrastructure 169 more facts Therefore your subroutine could take the form of a single defrule in its own defmodule If that defrule has the auto focus property so that it fires as soon as it s activated regardless of what other rules may be on the agenda and uses return on its RHS to resume the previous module focus as soon as it ran then the interview module can call it as a subroutine just by asserting a fact to activate it The trigger fact looks like ask id where id is a question identifier Such a rule can look like this defmodule ask defrule ask ask question by id Ask a question and assert the answer declare auto focus TRUE If there is a question with ident id MAIN quest
276. ed kinds of data Variable FuncallValue FactIDValue and LongValue are the four of most 246 15 4 15 4 1 15 4 2 CHAPTER 15 Extending the Jess language interest to us When you want to create a value to represent a variable a function call a fact or a Java long you must use the appropriate subclass Refer to the API documentation for more information about these classes Beyond simple examples We ve covered all that you need to know to write many useful Jess extension func tions In this section we ll look at some subtle techniques that can make your Userfunctions even more powerful Holding state A Userfunction is really just an ordinary Java class and so it can do things that ordinary Java classes do In particular it can contain member variables Many of Jess s built in functions contain member variables that hold data for the functions to operate on For instance the class that implements the bag function which lets you create modify and maintain collections of objects includes a Hashtable member variable that is used to store all the collections you create When you load a Userfunction into Jess only a single object of your class is created that single object is used every time the Userfunction is called so the values of those mem ber variables are preserved Note that the Rete class has a method findUserfunction that returns the object that implements a given function If you know the class name of t
277. ee section 7 6 for information about modules Asserting the fact clean up order order will then clean up the order number order defmodule CLEANUP defrule CLEANUP initialize order 1 declare auto focus TRUE MAIN initialize order number Pitem lt line item order number number gt retract item defrule CLEANUP initialize order 2 declare auto focus TRUE 304 17 9 CHAPTER 17 Jess on the Web MAIN initialize order number rec lt recommend order number number gt retract rec defrule CLEANUP initialize order 3 declare auto focus TRUE Pinit lt MAIN initialize order number not line item order number number not recommend order number number gt retract init defrule CLEANUP clean up order declare auto focus TRUE clean lt MAIN clean up order number Porder lt order order number number gt assert initialize order number retract clean order The rule initialize order 1 deletes all the line item facts when it sees an initialize order fact and initialize order 2 deletes all the recommend facts initialize order 3 deletes the initialize order fact when no more line items or recommendations are left Finally clean up order serves as the entry point it retracts the order fact and then runs the other rules by asserting the initialize order fact TIP It s impor
278. eed to learn how to embed the Jess rule engine into another application Part 6 TekMart com rule based applications for the Web G theability of rule engines to cope with incomplete information and unpredictable events rules and the World Wide Web are a perfect match Rule based systems are a part of most major web sites where they are used for order processing user preferences data mining and recommendations In part 6 you ll develop a Recommendations Agentas part of an e commerce web site based on servlets and JavaServer Pages Whereas in each of the previous applications you ran Jess as a free standing application in this case you ll use Jess as a library embedded in the Tomcat serv let engine Therefore you need to learn more about Jess s Java APIs In addition the rules you ll write for this application will be the longest you ve written so far in fact they approach the size of the rule that solved Mrs Rosencrantz s word problem in chapter 1 Jess on the Web In this chapter you ll Look at alternative Java web architectures Collect knowledge for a smart e commerce application Write rules and queries for a web application 287 288 CHAPTER 17 Jess on the Web A sizeable and still growing fraction of all software written today is deployed on intranets or the Internet The rise of electronic commerce enterprise applica tions and web based entertainment is a visible indicator of this trend Rul
279. efer to applications that use the HTTP and HTTPS protocols either over corporate intranets or over the Internet as a whole In fact new methods are being developed all the time and many of these methods blur the traditional categories making it hard to use any absolute terms to describe what is possible on the Web For our purposes though web based soft ware can be divided into two broad categories fat client and thin client applications Fat client applications A fat client web application is most similar to a traditional client server application The code is divided more or less equally between a client desktop machine and a server remote machine Fat client solutions can automatically download and install the client software to the desktop machine or they may use fixed clients desktop applications installed using physical media or manual network down Java architectures for the Web 289 loads The client software might include an elaborate graphical interface as well as some fraction of the business logic for the application There are two ways to build Java based fat client solutions that can be automat ically downloaded to the client using an applet or using Java Web Start Applets The Java Applet API was included in the first release of the Java Developer s Kit JDK A Java applet is just a class that extends java applet Applet and provides implementations for one or more of the Applet methods init start stop an
280. efined routers and even create your own router types In the next chapter you ll use what you ve just learned to deploy the Recom mendations Agent you developed in chapter 17 as a web application Deploying web based applications In this chapter you ll Learn about the Tomcat servlet engine Develop serviets that embed Jess Write JavaServer Pages that use Jess Assemble the Recommendations Agent from these components 323 324 CHAPTER 19 Deploying web based applications In chapter 17 you developed the rule base for a Recommendations Agent The agent can recommend further items to a customer based on their past and present purchases much as a good sales clerk working on commission might do In this chapter you ll build a web application around the Recommendations Agent In the first few sections I ll briefly describe some of the APIs and tools you ll be using and then you ll begin to incrementally develop the application You re going to deploy this application using the Java 2 Enterprise Edition envi ronment J2EE is a standardized deployment platform for networked Java applica tions Although you ll use specific software to make the examples concrete in particular the Tomcat servlet engine the application you ll develop is completely general and can be installed in any J2EE compliant application server Using Jess for this application gives you the ability to use a rule engine while writing only por table
281. eight in gold so it is 156 CHAPTER 9 Collecting the knowledge important to seek one out and develop a good working relationship Most of knowledge engineering consists of extracting information from domain experts For the forms advisor application potential experts include accountants and Internal Revenue Service IRS workers An accountant can tell you what forms are required most often by her clients whereas an IRS employee may have statis tics on form usage by the whole U S population Both can help you understand the tax rules 9 2 3 Desk research Not all of your information should come from people When possible you should instead collect basic or rote knowledge from written materials so as not to waste other people s time Of course you can t believe everything you read make sure the experts you talk to would trust the resources you use Books and journals You might use two broad categories of written material paper publications and electronic ones With the explosion of the World Wide Web during the last decade the amount of electronic research material available has mushroomed Still scholarly books and periodicals have a significant advantage over most elec tronic publications They are usually peer reviewed In the peer review process material destined for publication is read and critiqued by impartial experts This process improves the accuracy and trustworthiness of the information In many scientific and e
282. electricity Asking questions with backward chaining 193 Let s see how this works After that we ll reexamine and extend the ask module from part 3 to meet the needs of the current project 12 3 Asking questions with backward chaining In the Tax Forms Advisor application an entire module named interview was dedicated to picking out questions to ask the user In general there was one rule in interview specific to each question the application might ask and another rule in the recommend module to use the answer Because that application included only a dozen or so possible questions this approach wasn t much of a problem but it obviously wouldn t scale well to a system with hundreds of ques tions You need something easier for the application you re writing now Ideally you d like to have just a few generic rules that somehow know how to ask the appropriate questions at the right time In this application when a diagnostic rule is partially matched and needs a MAIN answer fact to match it further you d like the appropriate question to be asked and the resulting answer fact to be asserted You ll use backward chaining so your new program will automatically know when to ask each question Recall from section 7 4 that in Jess backward chaining works by creating special trigger facts based on partially matched rules To make backward chaining work you first have to turn it on with the do backward chaining function then
283. emory chips are typical of the class of problems that your program will diagnose It won t be able to help with questions like Why can t I connect to the network and What is a General Protection Fault Finally let s restrict the domain to include only computers that won t boot This leaves out many hardware problems You won t try to diagnose nonfunctional network cards jammed CD ROM drives and other broken components not directly involved in booting the machine Writing the first rules A simple top level flowchart for the program is shown in figure 12 2 If the com puter makes no sound at all the presumed hypotheses are The computer isn t getting any electricity and The power supply is faulty You can write these two hypotheses as rules in the MAIN module as shown in listing 12 2 The rule not plugged in handles the case where the computer makes no sound and is not plugged in the rule power supply broken handles the case where the silent computer is plugged in The two versions of the first pattern in the two rules represent the two outcomes from the decision Js there any sound in the center of figure 12 2 192 CHAPTER 12 Writing the PC Repair Assistant Listing 12 2 Rules to diagnose a computer that makes no sound deffunction recommend action action Give final instructions to the user printout t I recommend that you action crlf defrule MAIN not plugged in declare auto focus TRUE
284. emperature 70 degrees This definition causes some difficulty if you change the set point from 70 degrees So in this new version you ll provide a method in the Thermometer class to allow the definitions of cold cool OK warm and hot to be redefined when the set point value changes You ll also pre pare the Thermometer class to determine the rate of change of the temperature and define three fuzzy terms to describe the rate of change decreasing zero and increasing The new Thermometer class is defined in listing 16 3 Controlling with fuzzy rules 277 package fuzzy import nrc fuzzy public class Thermometer extends BeanSupport implements Runnable private Hardware m_hardware private int m_floor private double m_oldReading private FuzzyValue m_fuzzyReading private FuzzyValue m_fuzzyReadingRateOfChange private static FuzzyVariable m_thermometerFV null private static FuzzyVariable m_thermometerRateOfChangeFV static try m_thermometerFV new FuzzyVariable thermometer 0 catch InvalidFuzzyVariableNameException nameEx NOTHING catch InvalidUODRangeException rangeEx NOTHING SetThermometerFVTerms 70 0 try m_thermometerRateOfChangeFV new FuzzyVariable thermometerRateofChange 10 catch InvalidFuzzyVariableNameException nameEx NOTHING catch InvalidUODRangeException rangeEx NOTHING try m_thermometerRateOfChangeFV addTerm decreasing new
285. en added or removed FACT jess Fact A fact has been asserted or retracted FOCUS The name of the module The module focus has changed HALT jess Rete A run has been executed RESET jess Rete A reset has been executed RUN jess Rete A run has been executed USERFUNCTION jess Userfunction A Userfunction has been added or removed USERFUNCTION_CALLED jess Userfunction A User function has been called USERPACKAGE jess Userpackage A Userpackage has been added or removed jess PrettyPrinter 419 These flags are all powers of two so you can combine them with the Java opera tor and test for individual types within a combined type with the amp operator Jess combines two more constants MODIFIED and REMOVED with the others to mean for instance that a fact was retracted FACT REMOVED public int getType Returns the type of event one of the constants in table B 1 public Object getObject Returns data associated with the event public Object getSource Returns the source of the event the Rete object that generated it B 6 jess JessListener This is the interface you implement if you want to receive JessEvents It works just like the event listeners in the java awt event package public void eventHappened JessEvent je throws JessException Handles events of event type This is the only handler method in this interface Your handler should test the getType method in the JessE
286. en instance of jess Rete The lock object is internal to the Rete instance so executeCommand is synchronized independently of any other concur rency controls This means that if your web application uses executeCommand to call the get new order number function see section 17 7 1 it needn t worry about interference between multiple threads Exchanging Java objects Probably the most frequently asked question on the Jess email discussion list is some variation on How can I send my Java variable to my Jess program This section provides one easy answer to this important question We ll discuss the fol lowing functions Rete store String Object Stores any object in a special hash table from Java Getting started with the Jess library 309 Rete store String Value Stores a jess Value in the table from Java Rete fetch String Retrieves a jess Value from Java store symbol value Stores a value in a hash table from Jess fetch symbol Retrieves a value from the hash table from Jess In the previous section you asserted a fact color red into Jess s working memory from Java What if instead of the symbol red you wanted the fact to hold a java awt Color object representing a particular shade of red It s not possible to construct a String argument for executeCommand that directly contains the Color object This general problem how to move objects in both directions between Jess and Java is solve
287. en the contents of the multislot Rete engine new Rete Deftemplate d new Deftemplate student A student engine d addSlot name Funcall NIL STRING d addMultiSlot courses Funcall NILLIST engine addDeftemplate d Fact f new Fact student engine f setSlotValue name new Value Fred Smith RU STRING ValueVector courses new ValueVector courses add new Value COMP 101 RU STRING courses add new Value HISTORY 202 RU STRING f setSlotValue courses new Value courses RU LIST Note that even if a multislot contains only one value you have to put it in a Val ueVector When you re programming in Java it s up to you to follow this rule The results of breaking it are undefined Ordered facts As you saw in section 6 4 ordered facts like shopping list bread milk jam point 20 10 power off are represented as if they were unordered facts with a single multislot named _ data like this shopping list __data bread milk jam point __data 20 10 power _ data off If you assert an ordered fact from the Jess language the deftemplate is created automatically if it doesn t already exist The same thing happens if you create the fact from Java The deftemplate is created during the call to the Fact object s constructor 314 18 2 3 18 3 CHAPTER 18 Embedding Jess in Java applications Fact f new Fact shopping list engine f setSlotValue _
288. ener handler In Java programming you add behaviors to components like buttons and text fields by writing event handlers An event handler is an implementation of a special interface with a member function that performs the desired behavior The inter face used for button clicks is java awt event ActionListener You register the event handlers you write by calling an appropriate method on the component for ActionListener the method is called addActionListener When some user input occurs the component creates an event object and passes it as an argument to the appropriate method of all the event handlers that are registered with it In the case of button clicks the event object is an instance of the class java awt event ActionEvent and the method in the ActionListener inter face is called actionPerformed The function read input in listing 13 2 is an event handler written in Jess The class jess awt ActionListener is an event adapter that lets you specify that a deffunction should be invoked when a Java ActionEvent occurs Jess supplies a whole family of these event adapters one for each event type defined by the Java APIs and they all are used the same way You can create one using the name of a deffunction and a jess Rete object see the next section as constructor argu ments Then you use the matching Java method on an AWT or Swing component to register the listener just as you would in Java When the component sends an Diagnosti
289. ent id not MAIN ask gt assert MAIN ask id return Checking the answers The ask module developed in section 10 6 was a simple general purpose mecha nism for asking questions For this application you need to extend it a little In particular whereas the previous application only asked questions with yes no and numeric answers this application needs to be able to validate three different kinds of answers Multiple choice multi One of a set of allowed answers Yes or no questions are a special case of this type Numeric Any integer value Text Any arbitrary text 12 4 1 Checking the answers 195 Furthermore looking ahead to chapter 13 this application will have a graphical interface The original version of the ask module depended on the fact that all answers were read from the console using the Jess read function read inter prets what it reads which guarantees that something that looks like an integer is held internally as a jess Value object of type RU INTEGER In the last applica tion then the Jess parser figured out if something was a number before the ask module ever saw it You can t make that assumption any longer because you re going to use a GUI not read to collect answers Some numbers may arrive as text and need to be parsed Modifying the ask module In general then you need to extend and enhance the ask module for this appli cation The first ex
290. entState vent return new Value State vent state RU ATOM class WriteVent implements Userfunction public String getName return set vent state public Value call ValueVector vv Context c throws JessException int vent vv get 1 intValue c Simulator sim InitSimulator getSimulator c String szState vv get 2 stringValue c try boolean state State vent szState sim setVentState vent state catch IllegalArgumentException iae throw new JessException set vent state Invalid state szState return Funcall TRU GI 15 5 5 15 6 Testing 253 The functions init simulator n heatpumps n floors which pump get hp state set hp state get vent state and set vent state considerably sim plify the rules you ll write in the next chapter Implementing a Userpackage Several of the functions you wrote in this section are implemented by classes that require constructor arguments and most of them depend on the existence of an InitSimulator object Therefore it wouldn t make sense to allow them to be installed into Jess one at a time Instead they can all be installed together as a sin gle package using the Userpackage interface The Userpackage for the hardware functions is straightforward package control import jess public class HardwareFunctions implements Userpackage public void add Rete engine engine addUserfunction new InitSimulator
291. er minor changes were required in the Java code to deal with the fact that you are no longer representing the vent state as open or closed 3 Tf you already have some knowledge of FuzzyJ and fuzzy rules in general you might notice that the FuzzyValues for the vent terms are created with singleton FuzzySets These specialized FuzzySets have a value of 1 0 at the single point specified and a value of 0 0 everywhere else Using this type of set in your fuzzy conclusions together with a defuzzification method that takes the weighted average of these singleton outputs you are implementing zero order Takagi Sugeno Kang TSK rules Zero order TSK rules have fuzzy inputs but they have crisp constant outputs See the FuzzyJ Toolkit User s Guide for more information about TSK rules 274 CHAPTER 16 Writing the rules fully slightly half mostly fully closed open open open open 1 0 ay a o pe g v g Figure 16 3 Probability chart for the 0 0 fuzzy terms corresponding to the five discrete settings 0 0 0 2 0 4 0 6 0 8 1 0 used for the vents With these changes made you can now consider the new set of rules needed to control the vents Because you are not changing the way you determine when the heat pump should be cooling heating or off you will essentially leave the rules that determine the heat pump state as they are However to allow the tempera ture to fluctuate close to the set point without turning off t
292. ere s a deffunction to return the next order number If there s no next order number fact this function creates one and returns the lowest possible order number which you ve arbitrarily specified as 1001 If there is such a fact this function increments the value modifies the fact and returns the unincre mented order number get new order number uses the order number query to get hold of the fact it needs deffunction get new order number bind it run query order number if not it hasNext then assert next order number value 1002 return 1001 else bind token it next bind fact token fact 1 bind number fact getSlotValue value modify fact value number 1 return number Before we move on to writing the Java code for the Recommendations Agent you need to assemble one more chunk of Jess code a module dedicated to initializing new orders Cleaning up Web users are notoriously unpredictable They ll click Stop to cancel transactions they ll click Back right in the middle of a series of screens As a result your web application needs a way to reinitialize an order that s been partially filled out Removing all the line items all the recommends and the order fact from work ing memory should do it because the web application will be able to reassert these facts based on other state information You can use a defmodule containing a few auto focus rules to do the cleanup s
293. erver when in fact the rule engine was in a separate process maybe even on a dedicated machine The good news is that unlike writing EJBs writing RMI servers is simple In this section you ll develop a flexible and powerful RMI based rule server in less than 100 lines of code There are basically four steps to writing an RMI server 1 Define a remote interface 2 Implement the remote interface 3 Write a server mainline 4 Run the rmic tool to create the stub and skeleton classes In the remainder of this section you ll develop an RMI based rule server that can be used with EJBs or in other situations The remote interfaces A remote interface not surprisingly defines the methods of an object that are available remotely For your Jess server you need two kinds of remote objects a 1 A good place to learn about RMI is from the RMI chapter of Sun s Java Tutorial see http java sun com docs books tutorial rmi index html 380 21 2 2 CHAPTER 21 Jess in the J2EE environment remote version of jess Rete the main class you want to interact with and a fac tory class that can create instances of your remote Rete A typical client contacts the factory to get a single instance of the rule engine for its own private use loads it with rules and then makes repeated method calls to it A remote interface has to extend java rmi Remote and all of its methods must be declared as throwing java rmi RemoteException In add
294. es The number one rule of operating a heat pump is that you can t quickly or repeat edly switch it between heating and cooling If you do the heat pump may be dam aged if condensation forms and then freezes inside the pump Industrial grade heat pumps can cost thousands of dollars so this can be an expensive mistake Your control system must respect this rule One easy way to do this is to use guard lines There will always be some target temperature for example 70 degrees If you defined your rules to heat or cool the building whenever the tem perature was different from precisely 70 degrees then your system would be con stantly switching between heating and cooling mode trying to get the temperature exactly right Instead you ll define a range of acceptable tempera tures delimited by upper and lower guard lines which may move up or down depending on conditions As long as the temperature is within the valid range the system will be turned off The guard lines therefore serve as a kind of buffer so the system can t be rapidly cycled between operating modes Table 16 2 is an update truth table that includes guard lines 258 16 1 4 16 2 CHAPTER 16 Writing the rules Table 16 2 Truth table for the HVAC Controller system including guard lines The two added col umns represent the regions between the set point and each guard line Too Cold Lower Guard Just Right Upper Guard Too Hot Off Start Heating Do
295. es 288 garbage collection 39 209 geek speak 154 General Protection Fault See GPF gensym 315 gensym 315 400 GET 338 get 67 400 getAttribute 338 getCause 317 getClass 90 getContext 316 get current module 124 400 get focus 400 get focus stack 127 400 getGlobalContext 308 getInitParameter 338 getInputRouter 320 getLineNumber 316 get member 70 215 400 get multithreaded io 400 getName Userfunction method 239 getOutputRouter 320 getOutStream 320 getProgramText 316 get reset globals 49 401 getRoutine 316 get salience evaluation 122 401 getWriter 325 global decision 268 global variable name 48 global variables 48 GNU make 324 goal seeking 116 goals 118 Google 318 government publications 156 GPF 191 graceful degradation 198 Graphical User Interface See GUI guard lines 257 260 GUI 37 188 195 204 229 295 in EJBs 377 H halt 191 198 215 304 401 hardware Java interface 225 hash table 141 HashMap 62 68 head 137 311 of a fact 78 of a list 45 heap size 39 heat leakage 227 heat pumps 223 257 258 operation 257 rapid cycling 260 Hello World 34 heuristics 18 24 Holmes Sherlock 117 HotSpot 38 39 HTML 181 290 291 295 324 325 327 changing 332 generated 332 HTML editor 291 HTML forms 325 335 336 HTTP 288 324 325 338 HTTPS 288 325 HttpServlet 290 HttpServletRequest 325 333 340 HttpServletResponse 325 HttpSession 340 HttpUnit 343 human resources 288 HVAC 222 HVAC Co
296. es written in the faster C language For example on many problems Jess outperforms CLIPS by a factor of 20 or more on the same hardware See http aaaprod gsfc nasa gov teas Jess JessUMBC sld025 htm and http www mail archive com jess users sandia gov msg03278 html for some benchmarks Summary 39 Jess s rule engine uses an improved form of a well known method called the Rete algorithm Rete is Latin for net to match rules against the working memory We ll look at the Rete algorithm in detail in chapter 8 The Rete algorithm explicitly trades space for speed so Jess can use a lot of memory Jess does con tain commands that let you sacrifice some performance to decrease memory usage Nevertheless Jess memory usage is not ridiculous and fairly large pro grams will fit easily into Java s default heap size of 64 megabytes Because Jess is a memory intensive application its performance is sensitive to the behavior of the Java garbage collector Recent JVMs from Sun feature an advanced Java runtime called HotSpot which includes a flexible configurable garbage collection GC subsystem The garbage collector is the part of the JVM that is responsible for finding and deleting unused objects Excellent articles on GC performance tuning are available at Sun s web site Although every Jess rule base is different in general Jess benefits if you adjust two parameters the heap size and the object nursery size For example o
297. esigned to provide an interface that other soft ware can use The term web service generally refers to an application that can be sent commands using XML based Simple Object Access Protocol SOAP mes sages Web services can be used to build web applications with regular user inter faces but they are also used as components of larger software systems Deploying an application as a web service lets people and companies around the world interact with it For example if company A deploys its order configuration application as a web service then company B s purchasing department can develop software to automatically purchase A s product JavaSoft offers the Java Web Ser vices Development Pack JWSDP an add on for the J2EE application server as a platform for web service development 1 Bruce A Tate Bitter Java Greenwich CT Manning 2002 292 CHAPTER 17 Jess on the Web Pros and cons of thin clients Thin client architectures have both good and bad points The advantages are as follows Little processing power is required on the client because much of the code executes on the server There are few or no client configuration problems Few requirements are placed on the client so compatibility isn t usually an issue Upgrades are easy because they happen at the server Here are the disadvantages a HTML user interfaces are not as interactive as Java GUIs JavaScript can help but at the cost of reintroducing po
298. esigning the application run your test code and run it often ideally after each change you make to the developing system Watch for changes that lead to test failures if you catch them right away it is easy to back them out while they are still fresh in your mind Appendix C describes one technique for automated testing of Jess language code Here are some test facts you can use to test ask question by id deffacts MAIN test facts question ident ql type string text What is your name question ident q2 type number text What is your estimated annual income ident q3 type number text How many dependents do you have question To test the rule you just need to reset assert an appropriate ask fact and run You can use watch all to help see what happens Jess gt batch taxes clp TRUE Jess gt reset TRUE Jess gt assert ask q2 lt Fact 4 gt Jess gt watch all TRUE Jess gt run FIRE 1 ask ask question by id f 2 f 4 What is your estimated annual income 15000 gt f 5 MAIN answer ident q2 text 15000 lt Focus ask lt f 4 MAIN ask q2 gt Focus MAIN lt Focus MAIN 1 When you enter run the rule ask question by id grabs the module focus and fires It asks the question and asserts a new fact holding the answer Then the focus returns to the original module MAIN in this case The subroutine then consists of one module one rule
299. essions is easier if you start from the inside and work your way out The innermost pattern here is bus color c which matches any bus fact The not around that matches only when there are no bus 7 3 4 Qualifying patterns with conditional elements 113 facts The car color c pattern matches any car facts and the and groups these two patterns together The entire and thus matches when there is a car but no bus of the same color Putting the and group into the outermost not means the whole pattern matches only when the and doesn t thus the whole thing can be translated as It is not true that for some color c there is a car of that color but no bus of that same color In the next section we ll look at another interesting way to combine not CEs into more complex groups The exists conditional element You can nest multiple not CEs to produce some interesting effects Two nots nested one inside the other are so useful that there s a shorthand notation the exists CE A pattern can be enclosed in a list with exists as the head An exists CE is true if there exist any facts that match the pattern and false otherwise which is precisely the meaning of two nested nots The exists CE is useful when you want a rule to fire only once although there may be many facts that could potentially activate it Jess gt defrule exists an honest man exists honest gt printout t There is at least one honest man crlf
300. esting books often forget to tell you that some tests although perhaps not as good as all possible tests are infinitely better than no tests Don t be discouraged by the sense that you can t completely test a rule based appli cation Remember that some testing is a lot better than none at all Recommending forms Writing the rules for the recommend module is a bit more challenging but also more fun Each of the rules examines the user and answer facts and recom mends one or more forms These rules directly encode the knowledge gleaned from the interview with a tax expert in chapter 9 178 CHAPTER 11 Writing the application The first set of rules in this module determines which variants of Form 1040 the user might need In keeping with the design description in chapter 9 your strategy will be to recommend all the forms the user might need if in doubt you will recommend the form This means the system may recommend that the user bring home several versions of the same form and that s OK Recall that the user can file Form 1040EZ if and only if Her income is less than 50 000 and She has no dependents and Her interest income is less than 400 You can directly translate these requirements into a rule as follows defmodule recommend defrule form 1040EZ user income i amp lt i 50000 dependents d amp eq d 0 answer ident interest text no gt assert recommendation form 1040EZ
301. et is basically a loaded set of rules ready to be executed The javax rules admin package includes two different RuleExecutionSet Provider classes RuleExecutionSetProvider itself includes methods for creat ing RuleExecutionSets from serializable objects and therefore can be used when the rule engine is located in a remote server the constructor arguments can be sent via RMI Another class called LocalRuleExecutionSetProvider includes additional methods to create RuleExecutionSets from nonserializable resources like java io Reader objects local files Given a RuleServiceProvider you can create a RuleExecutionSet from a local file rules xml like this RuleAdministrator admin serviceProvider getRuleAdministrator HashMap properties new HashMap properties put name My Rules properties put description A trivial rulebase FileReader reader new FileReader rules xml RuleExecutionSet ruleSet null try LocalRuleExecutionSetProvider lresp admin getLocalRuleExecutionSet Provider properties ruleSet lresp createRuleExecutionSet reader properties finally reader close You then register the RuleExecutionSet with the RuleAdministrator under a well known name At runtime you use the same name to create a RuleSession the RuleSession uses the named RuleExecutionSet This example uses the well known name rules admin registerRuleExecutionSet rules ruleSet properties
302. et of overloaded methods is called a set of overloaded methods usually all do the same thing and the overloading is just for the sake of convenience This is the case with java io PrintWriter println Sometimes however you may want to call a specific overloading of a method and circumstances may conspire to make this impossible For example if you pass the string TRUE to a Java method that is overloaded to take either a boolean or a String it is generally impossible to predict which overload Jess will choose In these cases you can usually resort to using an explicit wrapper class For example suppose that in this case you want to invoke the boolean overload but Jess calls the String one instead creating and passing a java lang Boolean object should fix the problem because Jess will automatically convert java lang Boolean to bool ean but not to String Sometimes calling Java methods isn t enough you may need to work directly with an object s member variables or the static member variables of a class Jess lets you do that too Accessing Java member data Some Java classes have public variables you may need to work with Sometimes these are objects like the familiar System out More commonly they are static constants like MAX_PRIORITY in java lang Thread or NORTH in java awt BorderLayout Of 2 See the Java Language Specification section 15 11 2 2 70 5 4 CHAPTER 5 Scripting Java with Jess co
303. etSystemResourceAsStream This allows files along the class path including files in JARs to be batched Jess uses the class loader that loaded the application object You can supply this special object as a constructor argument when you create a jess Rete object 394 APPENDIX A Jess functions bind lt variable gt lt expression gt Assigns the given value to the given variable creating the variable if necessary Returns the given value bit and lt integer gt Performs the bitwise AND of the arguments bit and 7 4 is 4 and is equivalent to the Java 7 amp 4 bit not lt integer gt Performs the bitwise NOT of the argument bit not 0 is 1 and is equivalent to the Java 0 bit or lt integer gt Performs the bitwise OR of the arguments bit or 2 4 is 6 and is equivalent to the Java 2 4 bload lt filename gt Decompresses and deserializes a file to restore the state of the current Rete object The argument is the name of a file previously produced by the bsave com mand I O routers are not restored from the file they retain their previous state Furthermore JessListener objects are not restored from the file again they are retained from their state prior to the bload If the filename contains special char acters it must be passed as a double quoted string bsave lt filename gt Dumps the engine in which it is called to the given filename argument in a format that can be read using bload
304. ex once the car begins to move This time the instructions are filled with many context sensitive decisions in fact there are more decisions than actions The control flow includes many loops and branches It would be next to impossible to write a single list of instructions covering every situation that might possibly arise considering that circumstances interact to constantly change the correct outcomes of each deci sion For example seeing a child s ball roll into the road is not a serious situation when the car isn t moving but it requires prompt and decisive action when the car is driving toward the bouncing ball If it s impossible to write a procedural program to control your robot how can you do it You can use declarative programming instead Declarative programming a different approach Much of the programming we do is procedural Rule based programming how ever is declarative In procedural programming the programmer tells the computer what to do how to do it and in what order Procedural programming is well suited for prob lems in which the inputs are well specified and for which a known set of steps can be carried out to solve the problem Mathematical computations for example are best written procedurally Note that I m using procedural in a slightly different way than is conventional Although object oriented programming is traditionally contrasted with the older procedural programming for the purposes of
305. example think about the ways Sherlock Holmes might solve a mystery He has a collection of evidence a handkerchief a fingerprint a dead body and can proceed in two different ways First he can draw conclusions from the avail able evidence adding his conclusions to the available information and continue until he s found a link between the evidence and the crime This is a forward chaining method Alternatively he can start from the circumstances of the crime form a hypothesis about how it happened and then search for clues that support this hypothesis This latter technique is an example of backward chaining Holmes generally used both techniques in combination to solve a mystery as a Jess programmer you ll do the same Jess supports both forward and backward chaining but Jess s version of back ward chaining is not transparent to the programmer You have to declare which kinds of facts can serve as backward chaining triggers and only specific rules you define can be used in backward chaining In truth Jess s reasoning engine is strictly a forward chaining engine and so backward chaining is effectively simu lated in terms of forward chaining rules Still the simulation is quite effective and Jess s backward chaining mechanism has many useful applications You will apply it in several of the systems you develop later in this book Backward chaining is often used as a way to pull required data into Jess s work ing memory from
306. examples With this book I set out to change that pattern The book you re holding is structured around a series of large fully developed and eminently practical examples of rule based program ming in Java This book can be used in several ways First it is a general introduction to rule based systems If you ve never encountered rule based systems before you ll want to read part 1 closely This first section of the book introduces the concepts behind rule based systems discusses their applications and shows some first examples of rule based programs written with Jess Part 1 also discusses what s involved in adopting a rule based solution at your company Although the pro gramming examples in later chapters use Jess as a vehicle the concepts presented will transfer to other rule engines easily Second this book is a programmer s manual for the Jess rule language Part 2 is part Jess language reference and part tutorial It first introduces you to the lan guage and how the language is integrated with Java Later chapters in this part discuss rules and working memory elements the data that rules operate on There s also a chapter describing some of the theory behind Jess and what makes it run fast xxvi ABOUT THIS BOOK xxvii Finally this is a cookbook for real rule based systems Parts 3 through 6 describe substantial realistic software systems in enough detail to teach you how to develop similar systems on your o
307. explanation str cat el e2 180 CHAPTER 11 Writing the application Listing 11 4 shows the remainder of the rules that handle special circumstances These rules are all in the recommend module Listing 11 4 Additional rules for special circumstances defrule form 2016EZ answer ident employee text yes answer ident reimbursed text no gt bind ex Unreimbursed employee expenses assert recommendation form 2016EZ explanation ex recommendation form 1040 explanation ex defrule form 2016 answer ident employee text yes answer ident reimbursed text yes gt bind ex Reimbursed employee expenses assert recommendation form 2016 explanation ex recommendation form 1040 explanation ex defrule form 3903 answer ident moving text yes gt bind ex Moving expenses assert recommendation form 3903 explanation ex recommendation form 1040 explanation ex defrule form 4868 answer ident on time text no gt assert recommendation form 4868 explanation Filing extension defrule form 8283 answer ident charity text yes gt bind ex Excess charitable contributions assert recommendation form 8283 explanation ex recommendation form 1040 explanation ex defrule form 8829 answer ident home office text yes gt bind ex Home office expenses assert recommend
308. ey don t always know how to explain the pro cedures they follow On the other hand many experts respond well to interviews and you can ask questions to fill in gaps in the expert s explanations If you are developing a rule based system that is strictly based on a proce dures manual or other document or if a human expert is not available then the 24 2 4 2 2 4 3 CHAPTER 2 What are rule based systems knowledge may be collected directly from written sources Collecting knowledge from books and other reference material has its own advantages and disadvan tages Although books are generally more organized than human experts they can be lacking in the kind of practical rules of thumb or heuristics that a practi tioner can supply On the other hand you rarely have scheduling and other logistical problems when attempting to read a book but these can be annoying obstacles when working with a human expert Another important aspect of knowledge engineering is organizing and struc turing knowledge A typical rule based system contains hundreds or thousands of rules Organizing the collected knowledge so that translation to rules will be straightforward is a challenging task for the knowledge engineer We ll discuss the knowledge engineering process in greater detail in chapter 9 Structuring data When all the knowledge has been collected the task of programming the system begins The best first step is to examine the knowledge
309. f available tools for working with it Web browsers can visualize 20 2 1 Rules and XML 361 it Commercial XML editors are available The XSLT pattern language can be used to write simple scripts that transform an XML document into a new XML docu ment in a different format or into a non XML document And many high quality parsers and APIs are available for working with XML from your Java programs Representing rules as XML makes sense for several reasons among them interoperability editability and searchability We ll visit these ideas in the following sec tions The availability of XML parsers and of the XSLT scripting language means that transforming rules from XML into the language used by any given rule engine is straightforward Transforming in the other direction however requires you to have a parser for the rule language and sometimes this is not a trivial thing to write or to obtain Therefore it makes the most sense to create and store rules as XML transforming them into the native format of a rule engine for execution Note that although XML has its advantages it s not very readable Complex logic represented in XML will always be harder to read than the same logic expressed in a dedicated rule language like the Jess language Interoperability Quite a few rule engines are available that are suitable for use in enterprise appli cations Jess is only one of them it s difficult for me not to inject my opinion that Je
310. f data a value object contains it returns one of the constants from the jess RU class see table 15 1 Many of these types are dis cussed in chapter 5 For each type table 15 1 also lists a method that tries to return a value appro priate for that type You ve already seen the intValue method which tries to return a Java int based on the data held in the value object In general a Value tries to automatically convert the data it holds to the type you request for instance if you call intValue on a Value holding 1 0 as an RU STRING the string will be parsed and the method will return 1 There are limits to this capabil ity of course If some data cannot be converted to the requested type intValue and all the other methods in table 14 1 will throw jess JessException 242 CHAPTER 15 Extending the Jess language Table 15 1 value types defined in the jess RU class and the methods used to retrieve each type Name Method Meaning ATOM atomValue Jess symbol STRING stringValue Jess string INTEGER intValue Jess integer VARIABLE variableValue Jess variable MULTIVARIABLE variableValue Jess multifield FACT factValue Jess fact FLOAT floatValue Jess float FUNCALL funcallValue Jess function call LIST listValue Jess list EXTERNAL_ADDRESS externalAddressValue Java object LONG longValue Java long 15 2 3 Resolving variable arguments If kill is called like t
311. f other threads to perform background duties like garbage collection In graphical programs user code typically runs in several threads Some of the code you write runs on the main thread like the setup code you ve written so far while other code runs on the event thread When a menu item is chosen or a button is clicked the event handlers that are invoked run on the event thread Sometimes this is unimportant but when code on the event thread and code on the main thread need to coordinate their activities you have to think clearly about what code is running where In the program you re developing the run function will be called on the main thread and so the code for all the rules that fire will execute on the main thread On the other hand you want the user to click the OK button on the GUI This action will trigger some code to run on the event thread and you want to have that affect the behavior of the rules The situation is depicted in figure 13 5 210 CHAPTER 13 Adding a graphical interface new JTextField 40 Rete runUntilHalt Main thread Event thread new JTextField 40 run until halt rule fires ask user user clicks OK actionPerformed read input assert user input Text Rete assert pattern match user input fact next rule fires Figure 13 5 Some of the code in a GUI application runs on the main thread and other code runs on the event thread The separate threads can co
312. f the rules can be activated Jess sees that price check could be activated if there were an appropriate price fact so it generates the trig ger need price waffles nil This matches part of the LHS of rule find price but this rule cannot be activated because there is no item number fact Jess there fore creates a need item number waffles nil request This matches the LHS of the rule ind item number which fires and asserts something like item number waffles 123 This fact activates find price which fires and asserts price waffles 1 99 thereby activating rule price check which then fires The price is reported Each of the rules has fired once The definitions of the functions fetch price from database and fetch number from database are not shown they are presumably written in Java using JDBC 120 7 5 CHAPTER 7 Writing rules in Jess Listing 7 4 Multilevel backward chaining Jess gt clear TRUE Jess gt do backward chaining item number TRUE Jess gt do backward chaining price TRUE Jess gt defrule price check do price check name price name price gt printout t Price of name is price crlf TRUE Jess gt defrule find price need price name item number name number bind price fetch price from database number assert price name price Jess gt defrule find item number need item number name bind number fetch number from database name as
313. fact getSlotValue name stringValue null gt lt TD gt lt TD gt lt fact getSlotValue part number stringValue null gt lt TD gt lt TD gt lt price gt lt TD gt lt TR gt lt gt lt TR gt lt TD gt lt TD gt lt TD gt lt B gt Total lt B gt lt TD gt lt TD gt lt total gt lt TD gt lt TR gt lt TABLE gt lt BODY gt lt HTML gt 19 13 Persistence One detail that may have occurred to you is that the web application won t run forever Sometimes it will be taken offline When it is the Rete object containing all the past purchases and recommendations will be lost This is bad because this kind of customer information is valuable How can you preserve it The Servlet class has a destroy method The servlet container calls destroy when it s being shut down or the web application is being taken offline You can implement destroy to save all this data so it can be restored later In this imple mentation the catalog is being read from a flat file it s actually stored in the same file as the rules as a deffacts so you ll save the order information in a flat file too If you look back at the definition of checkInitialized in section 19 8 1 you ll see that that method attempts to load facts from the file named in the ini tialization parameter factsfile Here s an implementation of destroy that writes the facts out to that same file public void destroy try Serv
314. facts The grocery list has been added to the working memory and assigned an index 1 called the fact id Every fact in the working memory has a fact id assigned in order starting with 0 so that a fact with a larger fact id was always added to working memory at a later time than a fact with a smaller fact id Here the fact id 0 was assigned to MAIN initial fact The fact id serves as a convenient way to refer to a fact when you want to modify it or remove it from working memory Jess also uses fact ids when it decides the order in which rules will be fired The qualifier MAIN that was prepended to the first field in the list called the fact s head see chapter 4 is the name of the current module A module is a named subset of the rules facts and other constructs you ve entered into Jess Modules are often used to divide rules and facts into functional groups and we ll study them in detail in chapter 7 For now it s enough to know that MAIN is the default module name and all the facts you define belong to it The assert function takes any number of facts as arguments and returns the fact id of the last fact asserted or FALSE if the last fact couldn t be asserted Typi cally this means the argument was a duplicate of a fact that was already in working memory Jess s working memory contains only unique facts Real collections both grow and shrink You can sell baseball cards as well as buy them Facts likewise come and go
315. fferent sym bols The best symbols consist of letters numbers underscores and hyphens hyphens are traditional word separators The following are all valid symbols foo first value contestant 1 _abc Jess gives special meaning to a few symbols they are like Java keywords The sym bol nil is like null in Java and TRUE and FALSE are Jess s Boolean values Note that case is significant TRUE and FALSE must be uppercase and nil must be lowercase Other symbols have special meanings only in certain contexts for example the symbol cr1f is translated into a newline when printed The jess Value class Internally all Jess values symbols numbers strings and others are represented by instances of the jess Value Java class You will work extensively with this class when you interface Jess and Java code jess Value objects are zmmutable once one is created the value it represents cannot be changed Every jess Value object knows the type of the datum it holds and you can fetch this type using the type member function The possible values are enumerated by a set of constants in the jess RU class RU stands for Rete Utilities Symbols are of type RU ATOM Numbers Jess s parser uses the Java parsing functions java lang Integer parseInt and java lang Double parseDouble to parse integer and floating point numbers respectively See the Java API documentation for those methods for a precise syn tax description The following are
316. frule WORK quit job defrule WORK quit job WORK job salary s HOME hobby income i amp gt i s 2 MAIN mortgage payment m amp lt m i gt call boss quit job _ Commands that accept the name of a construct as an argument like ppdefrule ppdeffacts and so on search for the named construct as described earlier Note that many of the commands that list constructs such as facts list deftemplates and rules accept a module name or as an optional argument If no argument is specified these commands operate on the current module Ifa module name is given they operate on the named module If is given they operate on all modules see appendix A for full descriptions of all Jess functions and the arguments they accept Module focus and execution control You ve learned how modules provide a kind of namespace facility allowing you to partition a rule base into manageable chunks You can also use modules to con trol execution In general although any Jess rule can be activated at any time only rules in the focus module will fire Note that the focus module is independent from the current module discussed earlier Initially the module MAIN has the focus so only rules in the MAIN module can fire Jess gt defmodule DRIVING TRUE Jess gt defrule get in car gt printout t Ready to go crlf TRUE Jess gt reset TRUE Jess gt run 0 In this example the rule
317. ftemplate person People in actuarial database slot name slot age slot gender TRUE Jess gt assert person age 34 name Bob Smith gender Male lt Fact 1 gt Jess gt facts f 0 MAIN initial fact f 1 MAIN person name Bob Smith age 34 gender Male For a total of 2 facts 6 3 2 Unordered facts 83 The name of the deftemplate person is used as the head of the facts You can define as many slots as you want in your deftemplate You can also include a short description as shown in the example When you assert an unordered fact you can specify the slots in any order and you don t have to include every slot When Jess displays an unordered fact for instance when you type the facts function the slots are always displayed in a standard order the order in which you defined them in the deftemplate If you omit any slots when you assert an unordered fact they re filled in using default values Default slot values Most relational databases support the idea of a don t care value in a particular column If you add a row to a table and don t specify a value for one or more col umns then this default value usually called NULL is used Jess lets you do the same thing with unordered facts When you assert unordered facts you can omit values for any number of slots and Jess will fill in the default value nil Jess gt assert person age 30 gender Female lt Fact 2
318. ften doing so is necessary because of the EJB program ming restrictions Jess uses multiple threads and synchronized methods and therefore technically can t be used inside an EJB container There are a number of possibilities for integrating external resources into EJB applications One is to ignore the restrictions and directly use the resource from your EJBs this approach while not recommended is widely used apparently with great success Although it might work when the application is run on the refer ence implementation or on a single processor using another application server it is likely to fail spectacularly if the application is later deployed on a server farm The application server s persistence and replication services almost certainly won t work if you break these rules A more robust way is to run the resource in a separate application and let the EJB communicate with it through some form of interprocess communication In the rest of this chapter we ll look at several approaches to doing this with Jess Finally there s the J2EE Connector Architecture JCA a standard Java API for connecting external resources to J2EE servers The JCA is rather complex and implementing a JCA connector is far beyond the scope of this book However using a connected resource can be simple Often it s just a matter of obtaining a reference to the resource from the Java Naming and Directory Interface JNDI Commercial application servers
319. g 326 deployment descriptor 327 destroy 350 example 325 HelloJess 329 installing 327 user sessions 339 ServletUnit 343 session 339 session beans 376 set 67 408 set point 259 set theory 265 setAttribute 338 setContentType 325 set current module 408 setDefaultCloseOperation 215 set factory 408 442 INDEX setgen 408 set member 70 408 set multithreaded io 409 set node index hash 142 409 set reset globals 49 205 258 409 set salience evaluation 122 409 setSlotValue 345 set strategy 121 409 setVariable 243 shadow facts 229 dynamic 92 static 91 sharp logical states 264 shell scripts 169 show deffacts 410 show deftemplates 87 410 show jess listeners 410 showMessageDialog 215 Simulator 258 SimulatorGUI 229 singleton 273 skeleton 379 slot qualifier 83 slots 82 89 135 default values 163 SOAP 291 sockets 410 sorting 181 sound 197 spanning set 177 sqrt 55 410 square root 55 stack traces 71 316 stakeholders 153 startup banner 173 stateful rule session 386 stateful session beans 376 stateless rule session 386 stateless session beans 376 static member variables See vari ables static member static variables See variables static store 309 stored procedures 302 str cat 179 410 str compare 181 410 str index 411 STRING 422 string constants 250 stringp 411 strings newlines in 179 str length 411 stub 379 stub class files 383 subroutines 167 169 170 subrules 110 subseq 411 subsetp 411
320. g information In this last bit of code from the Recommend servlet the choice is between rendering the recommendations in the recommend JSP or check ing out via the Purchase servlet based on whether the recommendations for order query returned results In either case you dispatch to the new URL If there were recommendations you store them in the HttpServletRequest first if result hasNext request setAttribute queryResult result dispatch request response recommend jsp else dispatch request response purchase catch JessException je throw new ServletException je The recommend JSP The JSP you need to render the recommendations looks a lot like the catalog JSP The only substantial difference in the output is the presence of a Because you bought column that links each recommendation back to some other product The code to produce the display is marginally different because each individual query result contains two facts instead of one a recommendation fact and a prod uct fact but the technique for accessing their slots is identical When the customer The recommend JSP 347 submits the form generated by the recommend JSP the list of selected items is sent to the Purchase servlet The text of recommend jsp is shown in listing 19 1 Listing 19 1 recommend jsp renders the output of the Recommendations Agent lt HTML gt lt page import jess gt lt jsp useB
321. g it For example a rule about the pas sage of time might occasionally need to update the age slot of all the persons in the working memory You can change the values in the slots of an unordered fact using the modify function Continuing the previous example Jess gt modify 0 age 23 lt Fact 0 gt Jess gt facts f 0 MAIN person name Jane Doe age 23 gender Female hobbies snowboarding restoring antiques For a total of 1 facts The first argument to modify is either a Fact object or a numeric fact id and all the other arguments are slot name value pairs that specify a new value for the named slot You can modify any number of slots in a fact with a single modify function You can use modify on multislots too Note that the fact id of a fact does not change when you use the modify func tion It s the same fact it just has new slot values If you ve used CLIPS you might notice that this is an important difference between CLIPS and Jess in CLIPS the fact id changes when you use modify This simple property is extremely useful A slot of one fact can hold the fact id of another fact as slot data and in this way you can build structures of related facts Because the fact id of a fact is constant the relationships won t be broken if the facts are modified Copying facts with duplicate If you know that John Doe is the same age as Jane and likes the same things but is male instead of female you can create
322. g rules or writing out completed rules because it can use existing XML parsing and writing librar ies to do the hard work We ll discuss XML based rule editors in section 20 5 Storage and retrieval One important use for rule engines in an enterprise environment is to implement business rules Business rules are stated procedures that a business follows to accomplish its work A business rule is like a policy but more concrete It specifies a situation and the action that must be taken in that situation obviously a great match for the capabilities of a forward chaining rule engine Many businesses have a collection of numbered or otherwise indexed business rules The collec tion is constantly modified as rules are added removed or changed If a rule engine is being used to carry out business rules then there must be some way to identify the rule or rules that implement a given business rule If a whole category of rules is to be removed or temporarily suspended then the ability to search rules in a flexible way becomes important Rules represented as XML are ideal for this kind of environment If the rules are stored as one large XML document then XML based tools can be used to 20 3 20 3 1 XML based rule representations 363 search for specific rules If individual rules are stored in their own XML files then these textual snippets can be stored in a relational database and searched using the database s own mechanisms In eit
323. g them Humans learn in action An essential part of an In Action guide is that it is example driven It encourages the reader to try things out to play with new code and explore new ideas There is another more mundane reason for the title of this book our readers are busy They use books to do a job or to solve a problem They need books that allow them to jump in and jump out easily and learn just what they want just when they want it They need books that aid them in action The books in this series are designed for such readers XXX about the cover illustration The figure on the cover of Jess in Action is a Muger del Xeque a sheik s wife The illustration is taken from a Spanish compendium of regional dress customs first published in Madrid in 1799 A sheik was the head of an Arab clan or tribe and the richness of his wife s robes and jewelry would be considered a testament to his authority and wealth The book s title page states Coleccion general de los Trages que usan actualmente todas las Nacionas del Mundo desubierto dibujados y grabados con la mayor exactitud por R M V A R Obra muy util y en special para los que tienen la del viajero universal which we translate as literally as possible thus General collection of costumes currently used in the nations of the known world designed and printed with great exactitude by R M V A R This work is very useful especially for those who hold themselves to be unive
324. ger gt Deletes the specified range from a list The first integer expression is the one based index of the first element to remove the second is the one based index of the last element to remove div lt number gt lt number gt Returns the first argument divided by all subsequent arguments using integer division do backward chaining lt lexeme gt Marks a template as being eligible for backward chaining If the template is unor dered that is if it is explicitly defined with deftemplate or defclass then it must be defined before calling do backward chaining In addition this function must be called before defining any rules that use the template duplicate lt fact id gt lt symbol gt lt expression gt Makes a copy of the given fact with the values of slots modified as indicated by the symbol expression pairs The fact id must refer to an unordered fact Each list is taken as the name of a slot in this fact and a new value to assign to the slot A new fact is asserted which is similar to the given fact but which has the specified slots replaced with new values The fact id of the new fact is returned It is an error to call duplicate on a shadow fact e Returns the transcendental number e 398 APPENDIX A Jess functions engine Returns the jess Rete object in which the function is called eq lt expression gt lt expression gt Returns TRUE if the first argument is equal in type and value
325. ggests that you can call a module like a sub routine You can call a module from a rule s RHS using focus and the module can return from the call using return Alternatively a module can act as a kind of background process or periodic task by using auto focus rules to wake itself and return to put itself back to sleep Both forward and backward chaining rules can only react to the contents of working memory They are passive in the sense that they wait for facts to appear before they can take action and then nothing happens until they get to the top of the agenda Sometimes you may want to take a more active stance and deliber ately search through working memory to find particular information Jess lets you do this easily as you ll see in the next section Searching working memory with defquery Jess s working memory is a lot like a relational database Each deftemplate is like a relation a table in the database The individual slots are the columns of the tables If you re familiar with industrial strength relational databases you re prob ably aware of database triggers which are a lot like forward chaining rules attached to a database that fire when the data matches some criterion You can apply rules to relational databases so it s a reasonable question to ask whether you can make queries against the working memory of a rule based system Jess offers the defquery construct which lets you do just that A defquery is a sp
326. ght contain dozens or even hundreds of these annotated flowcharts Due to obvious space limitations you ll see only a few sim ple ones here The problem domain Of course problems with desktop computers is a very broad domain and you need to narrow it down a lot before you can get started First let s specialize desk 12 2 1 The problem domain 191 top computers to x86 based desktop computers We won t consider Macintosh machines or any other architectures just those systems that are commonly referred to as PCs You can formalize this specialization in a rule defrule MAIN right architecture declare auto focus TRUE answer ident hardware text x86 gt recommend action consult a human expert halt You ll write the function recommend action later Note that this rule like many others in this system calls halt to terminate execution Although this isn t really necessary right now you ll see why you need to do this in chapter 13 when you add a graphical interface to this application The auto focus declaration ensures that as soon as the system has a recommendation to make it will be made even if some other module has the focus at the time You ll use this decla ration in all the rules that make recommendations To further restrict the domain the system you re writing diagnoses only hard ware problems not software issues Broken power supplies loose video cards and faulty m
327. ght encounter early in the pro cess would be in the Jess language defrule use ez form If filing status is single and filing status single user made less than 50000 income i amp lt i 50000 gt recommend the user file Form 1040EZ recommend 1040EZ Detailed comments like those shown here will help non technical people read and understand the rules if necessary Buy a stack of white index cards and write each potential rule on one side of an individual card Use pencil so you can make changes easily The cards are useful because they let you group the rules accord ing to function required inputs or other criteria When you have a stack of 100 cards or more the utility becomes obvious You can use the reverse sides of the cards to record issues regarding each rule This stack of cards might be the final product of knowledge engineering or the cards contents might be turned into a report The cards themselves are often the most useful format though After organizing the new knowledge on index cards you may see obvious gaps that require additional information Develop a new set of interview questions and 9 2 2 Introduction to knowledge engineering 153 meet with the expert again The appropriate number of iterations depends on the complexity of the system Knowledge engineering doesn t necessarily end when development begins After an initial version of a system is available the expert shou
328. gine c getEngine Simulator sim InitSimulator getSimulator c int nHeatPumps sim getNumberOfHeatPumps if m_name equals N_HEATPUMPS return new Value nHeatPumps RU INTEGER else return new Value nHeatPumps 3 RU INTEGER The CountDevices class retrieves its associated InitSimulator object from Jess to get access to the simulator once it has that the rest of the work is trivial One thing worth noticing about this class and the previous one is that the function names are all defined as static final string variables like InitSimulator NAME and CountDevices N_FLOORS Each of the function names is used several times in getName possibly as a constructor argument and as here sometimes in the pro gram logic for those functions that decide what to do based on their name It s important that every use of the name match exactly so it s good programming practice to define each name as a constant in one place Matching heat pumps and floors Given the number of a floor which heat pump corresponds to it Heat pump 1 services floors 1 2 and 3 whereas heat pump 2 services floors 4 5 and 6 One way to compute this relationship is to subtract 1 from the floor number perform integer division by 3 and add 1 to the result This is not a slow calculation but it s a bit messy so you implement it in Java package control import jess class WhichPump implements Userfunction public String getName
329. global objects like socket factories and System out Load native libraries a Use static variables Some of these limitations are trivial who would expect to be able to display a GUI from a server side component but others can cause headaches Not being able to use synchronized methods is one of these most nontrivial code uses at least some synchronization Jess included The restrictions on accessing files loading native libraries and especially using static data can severely complicate many applications To be fair some of the restrictions can be safely ignored if you understand them For example if you know that an application won t use replication ser vices that is the J2EE container will always consist of just one process then you can ignore the admonition about static data Do you need to use EJBs If servlets and JSPs are the bread and butter of Java programming on the server EJBs must be the caviar By writing your program to the EJB specification you get a wealth of industrial strength qualities and services for free scalability availabil ity object persistence transaction management and more But there s no such thing as a free lunch and with EJBs as in life you have to pay for what you get First there are those restrictions If your application meets the definition of a traditional EJB application a database driven application that doesn t need to do any substantial computation then the restrictio
330. graphical kiosk then you may only need to change this one part of the system Simple text based I O Recall from section 3 1 4 Jess s printout function which you can use to print to standard output This function can accept any number of arguments and can per form rudimentary formatting you can control where newlines go by using the special symbol crlf as an argument There is also a function read that reads a single input token from standard input returning what it reads This suggests you can put these two functions together into a deffunction that emits a prompt and reads the response like this deffunction ask user question Ask a question and return the answer printout t question return read You should test this function to make sure it works assuming you ve entered the code for ask user in the file taxes clp Jess gt batch taxes clp TRUE Jess gt ask user What is the answer What is the answer 42 42 I entered 42 as the answer and the function returned 42 it appears to work fine So far ask user doesn t do any error checking You d like it to only accept answers appropriate to the given question for example only yes or no or only a number You need another function one that can check the form of an answer Here s one deffunction is of type answer type Check that the answer has the right form if eq type yes no then return or eq Panswer yes eq Panswer no else if eq
331. gt bind x The value The value To see the value of a variable at the Jess gt prompt you can simply type the vari able s name Jess gt bind a 2 2 4 Jess gt a 4 You can use a variable anywhere a value is expected For example variables can be passed as arguments in function calls Jess gt a 2 6 Global variables Any variables you create at the Jess gt prompt or at the top level of any Jess lan guage program are cleared whenever the reset command is issued Because reset is an important function it is used to reinitialize the working memory of Jess s rule engine this makes these top level variables somewhat transient They are fine for scratch variables while you are working interactively at the Jess gt prompt but they are not persistent global variables in the normal sense of the word To create global variables that are not destroyed by reset you can use the defglobal construct defglobal lt global variable name gt lt value gt This syntax diagram says that you can declare one or more defglobals in a single defglobal construct Each declaration consists of a global variable name an equals sign and a value Global variable names must begin and end with Valid global variable names look like this a all values counter 4 2 4 Adding some structure 49 Aside from the special naming convention defglobals are similar to ordinary variables and can be used in the same way
332. gt lt xsl text gt lt xsl apply templates gt lt xsl text gt lt xsl text gt lt xsl template gt lt xsl template match pattern gt lt xsl if test binding gt lt xsl text gt lt xsl text gt lt xsl value of select binding gt lt xsl text gt amp lt lt xsl text gt lt xsl if gt lt xsl text gt lt xsl text gt lt xsl value of select name gt lt xsl apply templates select slot gt lt xsl text gt lt xsl text gt lt xsl template gt lt xsl template match slot gt lt xsl text gt lt xsl text gt lt xsl value of select name gt lt xsl for each select gt lt xsl if test position 1 gt lt xsl text gt amp amp lt xsl text gt lt xsl if gt lt xsl apply templates select gt lt xsl for each gt lt xsl text gt lt xsl text gt lt xsl template gt lt xsl template match slot function call gt lt xsl text gt lt xsl text gt lt xsl call template name funcall gt lt xsl template gt lt Rule right hand sides gt lt xsl template match rhs function call gt lt xsl text gt amp xA lt xsl text gt lt xsl call template name funcall gt lt xsl text gt lt xsl text gt lt xsl template gt 371 372 20 5 CHAPTER 20 Jess XML and the enterprise lt Function calls gt lt xsl template match function call gt lt xsl call template name funcall gt lt xsl template gt lt xsl template
333. h quality server side solutions for the Java platform In this book we ll always refer to this software as the J2EE reference implementation Note that the J2EE is the sub ject of many books I don t have room here to teach much about J2EE we re only interested in how rule engines fit into the J2EE environment I will not pro vide listings of the classes or methods in any of the J2EE APIs although here and there PII discuss a small part of one of these APIs to provide context for the dis cussion I assume that you have some experience with the J2EE but I ll refresh your memory as appropriate What does that stand for Yow re going to meet a bewildering collection of acronyms in this chapter and the next I ve collected them into table 20 1 for easy reference Note that this table doesn t include all the acronyms related to J2EE development only the ones I ve used in this book 360 20 2 CHAPTER 20 Jess XML and the enterprise Table 20 1 Acronyms used in our discussions of J2EE Acronym Definition Meaning DAML DARPA Agent Markup Language An XML based language for representing general knowledge DOM Document Object Model A standard representation of an XML document as a tree of objects DTD Document Type Definition A way of specifying the allowed contents of an XML document EJB Enterprise JavaBeans A component model for server side Java programming J2EE Java 2 Enterprise Edition A specific
334. hare information between the pieces of a web application and you re using them all The Rete object is stored in the ServletContext it is shared among all invo cations of all servlets and JSPs for the whole lifetime of the web application You put the login name and order number in the HttpSession object so these are available only within a single browser session You use the HttpServletRequest to pass data from a servlet to a JSP this data exists only during the chain of servlets and JSPs fired off in response to a single browser request Each way of communicating between servlets and JSPs has its purpose and by using them properly you can make your application easier to understand Invoking the JSP All that s left for Catalog doGet to do is catch any thrown exceptions and then if all is well forward the query results on to the JSP for rendering catch JessException je throw new ServletException je dispatch request response catalog jsp The ServletException class has a constructor that accepts another exception object as an argument which makes it easy to report errors from Jess If a servlet 19 8 6 The Catalog servlet 341 throws a ServletException Tomcat returns an error page that shows the stack trace and error message from both the ServletException and any nested excep tion object You can then see all this information in your browser In a real applica tion you d
335. has x hair assert isa x mammal Here s the same rule in RuleML lt imp label AnimalsRulel priority 10 gt lt _head gt lt conclusions gt lt assert gt lt fact gt lt atom gt lt _opr gt lt rel gt isa lt rel gt 364 20 3 2 CHAPTER 20 Jess XML and the enterprise lt _opr gt lt var gt x lt var gt lt ind gt mammal lt ind gt lt atom gt lt fact gt lt assert gt lt conclusions gt lt _head gt lt _body gt lt and gt lt fact gt lt atom gt lt _opr gt lt rel gt has lt rel gt lt _opr gt lt var gt x lt var gt lt ind gt hair lt ind gt lt atom gt lt fact gt lt and gt lt _body gt lt imp gt You ll notice right away that the RuleML version is much longer This is typical of most uses of XML the reason is simply that the XML contains more information The individual tags explicitly state the purpose of each component of the rule whereas in the Jess format that purpose is implicit in the syntax The explicit tags make translating the XML representation into other rule languages much easier the verbosity is just the price you have to pay DAML DAML http www daml org is the DARPA Agent Markup Language DAML is an ambitious project to create a language for describing general knowledge on the World Wide Web DAML contains specific primitives for expressing relation ships like implies and is a kind of The most important aspect of DAML is
336. hat func tion you can retrieve the object cast it to the known type and retrieve any state it holds This is a convenient way to write a collection of cooperating functions that must share state Recall that a single Java program might contain multiple instances of the jess Rete class If you install a separate instance of your Userfunction into each one then each one will maintain its own state separately When you write the Userfunctions for the HVAC Controller system the classes will include a member variable to point to the Hardware object on which they operate The multiple classes will use indUserfunction to share this state Multiple personalities Consider this trivial Userfunction class include jess public class AnyString implements Userfunction private String m_name public AnyString String name m name name public String getName return m_name 15 4 3 15 4 4 Beyond simple examples 247 public Value call ValueVector vv Context c return new Value m_name RU STRING It can be constructed using any string when it is registered with Jess that string will be its name It will return the same string as its result You won t be able to use load function to load this class into Jess because load function won t let you specify the constructor argument instead you can call the addUserfunction method in the Rete class from Java Although this function isn t useful as is it illustrates an impor
337. he customer if they d like to buy a second videotape or DVD 17 5 5 Writing the rules 299 whenever they pick one out themselves Although this version of the Recommen dations Agent doesn t do it a more fleshed out version could offer a discount on purchases of multiples Buy one get a second for half price These first three patterns identify all pairs of videotapes and DVDs the third pattern ensures that you consider only pairs of two different items defrule recommend more media If customer buys a disk or tape recommend a random other item of the same category p1 lt product part number part1 category c amp dvd disk videotape name name p2 lt product part number part2 category c test neq pl p2 The test pattern compares the fact ids of the matched facts for inequality this is acommon idiom in Jess The line item pattern identifies orders that include one member of a pair of media and then the negated pattern eliminates orders that already include a dis cretionary media recommendation for this media type line item order number order part number part1 not recommend order number order type sym cat discretionary c Because the customer is buying the first member of a pair of items this rule rec ommends the second member gt assert recommend order number order because name part number part2 type sym cat discretionary c Mor
338. he flowchart to each action box to see how to define each rule For example the Check keyboard and motherboard action is 198 12 5 2 CHAPTER 12 Writing the PC Repair Assistant reached by passing through the seek does beep and how many beeps deci sions so it looks like this defrule MAIN motherboard or keyboard declare auto focus TRUE answer ident sound text yes answer ident seek text no answer ident does beep text yes answer ident how many beeps text t test gt integer t 3 gt recommend action halt check keyboard and motherboard Note that you have to use the integer function to convert the answer to a number before checking if it s greater than or equal to three You use a test conditional element instead of putting the test on t directly into the last answer pattern due to a limitation imposed by the backward chaining machinery Patterns that match backward chaining reactive patterns can only contain simple tests function calls aren t allowed The check ram rule is similar but with a twist Besides recommending an action this rule asserts a check fact and does not call halt defrule MAIN check ram declare auto focus TRUE answer ident sound text yes answer ident seek text no answer ident does beep text yes answer ident how many beeps text t test lt integer t 3 gt assert check loose ram recom
339. he heat pump you need to make a minor change to the rules floor hot enough and floor cool enough In these two rules rather than turn off the heat pump as soon as all the floors have passed the set point you allow a 0 25 degree overshoot or undershoot of the set point before making the decision to turn off the heat pump Why do this and why 0 25 degrees Consider the following situation If the heat pump is cooling then the situation is normally that the outside temperature is above the set point temperature If the temperature cools to the set point and you shut the heat pump down as soon as this happens then the heat pump cannot be turned on again until it reaches the guard limit in this situation the set point tempera ture 2 0 degrees In the non fuzzy case where you can only have the cooling either fully on or fully off this makes some sense But now that you can change the amount of cooling by controlling the vents you want to give the system an opportunity to adjust the amount of cooling Allowing an overshoot or under shoot of the set point gives the system an opportunity to find a vent opening that provides just enough cooling to match the heat that is warming the system The 0 25 value was an initial guess and it turned out to provide quite good behavior A bit of experimentation might allow you to reduce it thus keeping the guard tem perature far enough away to avoid too rapid cycling from cooling to heating you Controlling
340. he problem Joe turns out to be the mysterious man in the blue pants and Fred the tasteless golfer in the orange ones will tee off first What would happen if some of the information was missing For example sup pose you didn t know that Joe was second at the tee how would this affect the results Let s give it a try If you change Joe s section of the program like so We don t know anything about Joe really position of Joe is p2 amp p1 pants color of Joe is c2 amp c1 1 1 1 CHAPTER 1 Rules to the rescue and then run it you get the following result Fred 3 orange Joe 4 blue Bob 1 plaid Tom 2 red Fred 2 orange Joe 4 blue Bob 1 plaid Tom 3 red Fred 2 orange Joe 1 blue Bob 4 plaid Tom 3 red Fred 1 orange Joe 2 blue Bob 4 plaid Tom 3 red Fred 1 orange Joe 3 blue Bob 4 plaid Tom 2 red Fred 1 orange Joe 4 blue Bob 3 plaid Tom 2 red Now there are six different solutions and Jess finds and reports them all This is another strength of rule based programming Rule based systems can degrade gracefully in the presence of incomplete information You didn t have to build this quality into the program it s just there Beyond logic puzzles When you wake up from the recurring math class nightmare you may be relieved to think that solving logic puzzles is a far cry from your normal duties as a pro grammer But that s really not so Ill defined problems like this logic puzzl
341. her case XML reporting tools can be used to display the rules found by a query Obviously the whole package we ve just described is greater than the sum of the parts If rules are developed in XML and stored in a searchable fashion then they can be edited transformed into executable form and deployed on demand In the rest of this chapter we ll flesh out these ideas by looking at XML based rule formats and developing an XML based rule editor XML based rule representations In the previous section we looked at reasons for wanting to store rules in an XML format In this section we ll address the question of exactly what format to use You have several emerging standards to choose from or you may define your own format Let s look at each of the alternatives in turn RuleML The RuleML project http www dfki uni kl de ruleml is defining a standard representation for rules of all kinds The proposed RuleML standard is very broad in its reach It not only covers forward and backward chaining rules of the sort appropriate for rule engines like Jess but also transformation and mapping rules and rules that define web services The following example is by Said Tabet who has written an XSLT stylesheet that converts a subset of RuleML into Jess rules http www dfki uni kl de ruleml jess Here s a simple Jess rule that says If an animal has hair then it isa mammal defrule AnimalsRulel declare salience 10
342. hes only facts that include that value In the following example although both facts have the head letters only the one with slot data that exactly matches the pattern activates the rule Jess gt clear TRUE Jess gt defrule literal values letters b c gt TRUE Jess gt watch activations TRUE Jess gt assert letters b d This doesn t activate the rule lt Fact 0 gt Jess gt assert letters b c but this does gt Activation MAIN literal values f 1 lt Fact 1 gt Remember that an ordered fact is implemented as an unordered fact with a single multislot named __data a multislot you ll recall can hold any number of items You could therefore write the previous rule as Jess gt assert letters b c lt Fact 0 gt Jess gt defrule literal values letters __data b c gt TRUE and it would behave the same way I asserted a letters fact first to emphasize that Jess only defines the implicit deftemplate for letters when it sees an ordered letters fact the rule won t be parsed correctly until this deftemplate 102 7 2 2 CHAPTER 7 Writing rules in Jess exists It s important to keep this relationship in mind as you read this chapter remember that everything that applies to ordered facts applies equally well to the multislots of unordered facts The same goes for the regular slots of unordered facts with the restriction that they can hold only one value at a time
343. his Jess gt kill errorCode then the type of the argument Value object is RU VARIABLE If you call intValue on such a Value the variable is automatically resolved and the return value is the value of the variable This is why intValue and friends require a jess Context as an argument Context objects contain information about the situation in which your function was called including a pointer to the Rete object and more rele vant here the current values of all the variables that have been defined In many cases the fact that each argument might be a variable is irrelevant variables are resolved automatically But because looking up the value of a variable is a more expensive operation than just fetching the data from a Value if you re going to reference a Value more than once in your Userfunction it may be more efficient to separate the value resolution from the data access using the resolveValue method You can call resolveValue on any Value For Value objects of most types resolveValue is a very inexpensive operation that returns the Value object itself But for some types it returns a new real Value For RU VARIABLE and RU MULTIVARIABLE types this is the value of the variable The 15 2 4 Handling arguments 243 following admittedly contrived example computes the fractional part of a real number it uses resolveValue to avoid looking up the value ofits argument twice public Value call ValueVector vv C
344. hoices Numeric A text field that accepts numeric input You ve essentially already done this one Once again the ask module can be modified to accommodate this latest require ment You can change the ask user function so that it sets up the appropriate input component for each question based on the type of question being asked For questions with numeric answers you use the existing JTextField and for multiple choice questions you use a new JComboBox Setting up the JComboBox and its associated OK button is similar to setting up the JText Field After the components are created the JButton needs an event handler to assert the selected item as an answer fact defglobal acombo new JComboBox create yes no defglobal acombo ok new JButton OK deffunction combo input EVENT An event handler for the combo box assert ask user input sym cat acombo getSelectedItem bind handler new jess awt ActionListener combo input engine acombo ok addActionListener handler 214 13 6 CHAPTER 13 Adding a graphical interface Diagnostic Assistant Please wait Figure 13 6 The multiple choice input panel displays a combo box and a button The choices in the combo box are changed based on the question being displayed One interesting bit about this code snippet is the call to the JcomboBox construc tor The constructor expects an array of Object as an argumen
345. hose used in relational data bases to make searching the working memory a very fast operation 22 2 3 4 2 3 5 CHAPTER 2 What are rule based systems It s up to the designer of the rule engine to decide what kinds of things can be stored in working memory Some working memories can hold only objects of a specific type and others can include for example Java objects The pattern matcher Your inference engine has to decide what rules to fire and when The purpose of the pattern matcher is to decide which rules apply given the current contents of the working memory In general this is a hard problem If the working memory contains thousands of facts and each rule has two or three premises the pattern matcher might need to search through millions of combinations of facts to find those combinations that satisfy rules Fortunately a lot of research has been done in this area and very efficient ways of approaching the problem have been found Still for most rule based programs pattern matching is the most expen sive part of the process Beginning rule programmers often overlook this fact expecting the procedural right hand sides of their rules to represent all the com putational effort in their program The solution to Mrs Rosencrantz s problem involved lots of pattern matching and no procedural code at all except to print a report at the end Often the pattern matching technique used by a particular rule engine will affec
346. ial purposes Jess has evolved quite a bit since its original introduction in 1997 largely in response to feedback from a global user community I ve enjoyed working on Jess the whole time and look forward to its continuing evolution in the future acknowledgments Writing a book is a huge project This is my second book and somehow I thought it would be easier this time around It wasn t The original four month estimate to write the manuscript has stretched out into much more than a year I m very happy with the results though Writing a book about a subject so near and dear as Jess is to me is a dodgy business I think I ve steered clear of the minefields of self indulgence and created something that will be useful to everyone interested in rule based software Writing a book is such a huge project in any event that no one does it alone least of all me I ve had help from many kind generous and talented people dur ing the whole time this book was being developed One standout has been Bob Orchard of Canada s National Research Council Bob is the author of the Fuzzy toolkit and the FuzzyJess extension that adds fuzzy logic to Jess He s been an active member of the Jess community for years He gen erously contributed the essay in chapter 16 showing how to apply the principles of fuzzy logic to the HVAC Controller example He also served as both a technical reviewer and a technical proofreader for this book and provi
347. ible to get the same effect using a single call to the clear function as you ll see in the next section Clearing and initializing working memory As you work interactively in Jess the working memory tends to become full of bits and pieces of information that are no longer relevant When this happens you might want to clean things up to make it easier to see what you re doing Similarly a running program might want to periodically start from a known state for instance when a particular customer signs on to your e commerce site the pro gram should have on hand only the information relevant to that customer You can remove all the facts from working memory using the clear function However clear goes beyond just erasing working memory it also erases all vari ables rules and deffunctions from Jess Because clear essentially deletes the entire active program it s not used very often generally only if you re in an interactive session and want to start from a clean slate To restore the initial state of an application without erasing it completely you can use reset The reset function puts working memory into a known state By default working memory is empty except for the special fact MAIN initial fact as you ve seen in the earlier examples This initial fact is special because Jess uses it internally Many rules implicitly expect it to be there and won t work correctly without it Before you use the working memory it s
348. icates that advice should be removed from all functions Jess functions 413 undefinstance lt object gt If the object currently has a shadow fact removes the shadow fact from working memory Furthermore if the object has a PropertyChangeListener installed this is removed as well If the argument is this is done for all definstances undefrule lt lexeme gt Deletes a rule Removes the named rule from the Rete network and returns TRUE if the rule existed This rule will never fire again If the rule has subrules gener ated by compiling the or conditional element they are removed as well union lt list gt Returns a new list consisting of the union of all its list arguments that is of all the elements that appear in any of the arguments with duplicates removed unwatch all rules compilations activations facts focus Causes trace output to not be printed for the given indicator See watch upcase lt lexeme gt Converts lowercase characters in a string or symbol to uppercase Returns the argument as an all uppercase string view Displays a live snapshot of the Rete network in a graphical window You can double click on individual nodes in the network to get information about them watch all rules compilations activations facts focus Produces debug output when specific events happen in Jess depending on the argument Any number of different watches can be active simultaneousl
349. ication the customer may be the postal service No one at the post office will be able to supply much domain knowledge but they will be able to describe the problem well enough Luckily it s obvious in this case who the domain expert should be a tax accountant An accountant knows better than anyone else which tax forms people need under various circumstances The cus tomer should be willing to pay for some of an accountant s time or perhaps pro vide access to their own accountants Users The users are the people who will interact with the system on a day to day basis Like the customers the users may or may not know much about the domain in which the system works A particular category of user the expert user knows the domain very well Expert users are people who will use your system to automate tasks they already know how to do They are often the best kind of interviewee to work with because they understand the problem and simultaneously know how they want the system to react The users for the forms advisor are not expert users they are just people who wander in to the post office to pick up tax forms This kind of user isn t particu larly useful to interview for knowledge engineering purposes however it can be useful to talk to naive users about things like user interface issues Experts A domain expert is someone who has technical knowledge in the relevant problem area for your system A good domain expert is worth her w
350. ields corresponding to the table s columns When you assert an unordered fact you can specify the slots in any order hence the name unordered Unordered facts are the most common kind of fact and a good choice for most situations An unordered fact looks like this person name John Q Public age 34 height 5 10 weight 170 An ordered fact lacks the structure of named fields it is just a short flat list but ordered facts are convenient for simple bits of information that don t need struc ture All the facts we ve used as examples so far have been ordered facts If you used an ordered fact to represent the same person data it would look like this person John Q Public 34 5 10 170 Finally shadow facts are unordered facts that are linked to Java objects in the real world they provide a way to reason about events as they occur outside of Jess In the following sections we ll cover each of these categories of fact in detail Ca ross Working memory employee name John Smith title Supervisor num 1 employee name Sue Jones Figure 6 2 Working memory is something like a relational database title Engineer num 2 82 6 3 6 3 1 CHAPTER 6 Representing facts in Jess Unordered facts A table in a relational database has a name and a set of named columns Each row of data in a table provides a value for each of the columns
351. if a mode string is given it may be opened for reading only r writing only w or appending a If the filename contains special characters it must be passed as a double quoted string or lt expression gt Returns TRUE if any of the arguments evaluate to a non FALSE value otherwise returns FALSE pi Returns the irrational number pi pop focus Removes the top module from the focus stack and returns its name ppdeffacts lt lexeme gt Returns a pretty print rendering of a def facts as a string ppdeffunction lt lexeme gt Returns a pretty print representation of a def function as a string ppdefglobal lt symbol gt Returns a pretty print representation of a defglobal as a string ppdefquery lt symbol gt Returns a pretty print rendering of a defquery as a string ppdefrule lt symbol gt Returns a pretty print rendering of a defrule as a string ppdeftemplate lt symbol gt Returns a pretty print representation of a deftemplate as a string 406 APPENDIX A Jess functions printout lt lexeme gt lt expression gt Prints its second and subsequent arguments to the router named by the first argu ment which must be open for output No spaces are added between arguments The special symbol cr1f prints as a newline The special router name t can be used to signify standard output progn lt expression gt Evaluates each of its arguments in turn and returns the value of the las
352. ifferent sets of rules to control the vents rules that open the vent when a floor needs service and rules that close it when the floor s temperature is acceptable The only thing you need to decide is where the impor tant temperatures are Because you want the vents to help moderate the system you ll switch them based on the set point temperature Note that because a vent doesn t accomplish anything unless the heat pump is on you need to include the state of the heat pump in each rule Table 16 3 is a truth table for what to do when the heat pump for a floor is in cooling mode Table 16 3 The truth table for vent operation when the heat pump for a floor is in cooling mode The rows correspond to the current vent state Too Cold Lower Guard Just Right Upper Guard Too Hot Open Close Vent Close Vent Do Nothing Do Nothing Do Nothing Closed Do Nothing Do Nothing Do Nothing Open Vent Open Vent This truth table could lead to oscillations of the vents If the temperature were hovering right at the set point the vents might open and close rapidly If this oscil lation became a problem and in a real system it would because the vents would be noisy then you could establish a second set of guard lines closer together than the ones you re using for the heat pump The vents could then be opened or closed only when the temperature passed these guard lines We won t complicate the HVAC Controller system with this detai
353. il http www sendmail org delivers most of the email on the Internet It sports a famously cryptic rule based configuration language The following example is lifted verbatim from the sendmail web site It shows a sendmail rule that translates BITNET addresses like decvax user and research user into user decvax dec com and user research att com respectively translate UUCP addresses to SMTP ones LOCAL_RULE_3 UUCPSMTP decvax decvax dec com UUCPSMTP research research att com Microsoft Outlook Eudora Netscape Messenger OS X Mail and other popular mail clients include the ability to automatically sort messages into mailboxes according to the sender s address the recipient s address the subject and other characteristics In all cases the user programs the filter in a rule based language For these graphical mail clients the rule language is generally a pattern of check boxes and pop up menus in a mail filtering dialog box Programs like procmail http www procmail org and filter http www math fu berlin de guckes elm elm index html do this same kind of filter ing in batch mode on Unix systems I d personally perish under the weight of the hundreds of messages I receive each day without procmail to sort through and 10 1 2 2 CHAPTER 1 Rules to the rescue organize my mail Instead of dialog boxes these programs offer a simple textual language For example here are two of the rules
354. ill be necessary you can bet on that Web sites are redesigned frequently and if a Java programmer needs to sit down edit the code and recompile the servlets every time management wants to change the layout of the tables you ve got a big maintenance problem on your hands You may have heard of the model view controller MVC paradigm for pro gramming user interfaces In an MVC design the data or model is separated from the logic or controller and the GUI or view all in well separated modules If you broke the Catalog servlet into pieces this way you could make the web site much easier to maintain The controller logic would go in the Catalog servlet and the view would be built directly in HTML the data will be held in a Rete object so the model is already nicely encapsulated The HTML pages could be modified by web designers the servlets could be modified when necessary by Java programmers and maintaining the web site suddenly becomes much easier The only problem is that the HTML won t be static The page produced by the Catalog servlet will change based on many pieces of data For example the cus tomer s name will surely appear on the page The list of products and prices will change from day to day as well How can you write HTML pages that can change in response to data in your servlets The answer is to use JavaServer Pages as you ll see in the next section JavaServer Pages JavaServer Pages JSPs are HTML docu
355. immerSwitch into working memory Here you ll create one from Jess code but it could come from anywhere later in this book you ll learn 6 5 6 Shadow facts 91 many strategies for passing objects back and forth between Jess and Java code After you create a DimmerSwitch you use the definstance function to add it to the working memory Jess gt bind ds new DimmerSwitch lt External Address DimmerSwitch gt Jess gt definstance dimmer ds static lt Fact 0 gt Jess gt facts f 0 MAIN dimmer brightness 0 class lt External Address java lang Class gt OBJECT lt External Address DimmerSwitch gt For a total of 1 facts The first argument to definstance is the name of a template created by defclass and the second argument is a corresponding JavaBean We ll discuss the meaning of static in the next section As soon as you call the definstance function a shadow fact representing the Bean appears in the working memory Rules can react to this dimmer fact just as they can to any other fact in working memory If you imagine thousands of dimmer facts representing the brightness of every light fixture in a large office building you can see how letting Jess reason about DimmerSwitch objects could be useful Static vs dynamic shadow facts The working memory representation of a JavaBean can be either static changing infrequently like a snapshot of the properties at one point in time or dynamic changing autom
356. important to use the 80 CHAPTER 6 Representing facts in Jess reset function at least once while your program is starting up or at the begin ning of an interactive session You or your program can issue reset again when ever you want to reinitialize the working memory You can specify the initial contents of working memory in addition to MAIN initial fact using the deffacts construct The deffacts construct Typing separate assert functions for each of many facts is rather tedious It s also common to initialize the Jess engine using the reset function which clears the working memory and then immediately want to put the working memory into a known state containing a number of initial facts For example in an e commerce program the initial facts might include the product catalog To make this process easier Jess includes the deffacts construct A deffacts construct is a simply a named list of facts You can define any number of deffacts constructs The facts in all existing deffacts are asserted into the working mem ory whenever the reset function is issued Listing 6 1 demonstrates the operation of deffacts Jess gt clear TRUE Jess gt deffacts catalog Product catalog product 354 sticky notes 1 99 product 355 paper clips 0 99 product 356 blue pens 2 99 product 357 index cards 0 99 product 358 stapler 5 99 TRUE Jess facts For a total of 0 facts Jess gt reset TRUE Jess g
357. in my procmail configuration Put all messages mentioning Jess in their subject line into the IN jess users folder 30 Subject jess Mail IN jess users Send all messages from Out of Office Autoreply to dev null the UNIX Trash can 0 From Autoreply dev null Other programs exist that can automatically filter mail at a mail server to remove spam or to strip viruses from attachments Again these programs offer a wide range of features and interfaces generally programmed via rules Email handling is one common example of how rules can make life easier for individuals Now let s look at some applications of rules in the enterprise Product configuration When a complex customizable product like a computer is sold to a customer the seller must make sure the order corresponds to a functional system If a given peripheral requires a specific cable then the order must include that cable Like wise if the chassis can hold only two full height disk drives then the order better not include four of them For many kinds of custom manufactured goods hun dreds or thousands of these kinds of restrictions exist making order validation a difficult and painstaking process The XCON system and its predecessors developed at Digital Equipment Cor poration DEC are well known examples of using rule based systems in this application area The original XCON included 210 rules for validating orders for DEC hardware By
358. inciples You need to learn the organizing principles of the field The resources Once you understand the principles you need to know where to go to learn more The frontiers Every domain has its dark corners and dead ends You need to find out where the tough bits ambiguities and limits of human understand ing lie The knowledge engineer can use many potential sources of information to research these points Broadly though there are two interviews and desk research In the rest of this section we ll look at techniques for mining each of these infor mation sources to gather the four categories of information we just listed Where do you start When youre starting on a new knowledge engineering endeavor it can be diffi cult to decide what to do first Knowledge engineering is an iterative process You usually can t make a road map in advance instead you feel your way along adjust ing your course as you go As the saying goes though a journey of a thousand miles begins with a single step and taking that first step can be hard With most projects you should first talk to the customers the people who are paying you to write the system Find out what their needs are and what resources they can make available This isn t knowledge engineering per se but requirements engineering part of planning any software project But the customer might point you to particular sources of technical information and help you plan y
359. ing on the materials involved we ll arbitrarily choose the value 0 01 If you do this for each source of heat for each floor for each time step you ll have a reasonable simulation of the temperatures in an office building The code to do this is shown in listing 14 2 HOT and COLD represent the temperatures of the hot and cold air coming from the heat pump and m_outdoor is the variable temperature of the outside air The 228 CHAPTER 14 The reality connection big while loop looks at each floor in turn calculating the contribution of each of the factors listed earlier to the new temperature for that floor Simulator s con structor starts a background thread that periodically updates the temperature for each floor Listing 14 2 An implementation of get Temperature package control import java util Arrays public class Simulator implements Hardware private final double RATE 0 01 private final double HOT 100 COLD 50 private double m_temperature double m_outdoor 90 public Simulator int numberOfFloors VE oot m_temperature new double numberOfFloors Arrays fill m_temperature 70 new Thread this start LP wine public double getTemperature int sensor return m_temperature sensor 1 public void run while true for int i 0 i lt m_temperature length i double temp m_temperature i Heating and cooling and heat rising switch state i case HEATING temp
360. ing slot data 107 an arbitrary function and the slot data must match whatever the function returns When you re writing a return value constraint the function is preceded by an equals sign You can rewrite the previous example using a return value con straint like so item price x item price x 2 The return value constraint version is simpler because you don t need the variable y or the call to eq In fact pretty printing a rule containing a return value constraint always shows that Jess has transformed it into an equivalent predicate constraint using eq so the two forms are equivalent Which one to use is a matter of taste Pattern bindings To use retract modify or duplicate on a fact matched by the LHS of a rule you need to pass a handle to the fact to the RHS of the rule To do this you use a pat tern binding variable Jess gt defrule pattern binding fact lt a retract me gt retract fact A reference to the jess Fact object that activates this rule is bound to the vari able act when the rule is fired You can retrieve the name of a fact its integer ID and other useful data by call ing the Java member functions of the jess Fact class directly like this Jess gt defrule call fact methods fact lt initial fact gt printout t Name is call fact getName crlf printout t Id is call fact getFactId crlf TRUE Jess gt reset TRUE Jess gt run Name i
361. ining more terms for the temperature s rate of change improve the control What would the new set of rules look like How do the shape and position of the fuzzy sets you have used affect the control Change them and explore Perhaps you can devise some tech niques to determine the best definitions for these fuzzy sets You did not use the outside temperature in any of your reasoning perhaps it would be useful in dynamically modifying the sets to get better control When the system reaches a controlled state the valve positions are still being modified at each cycle perhaps by small amounts Can you think of some techniques to stop moving the valves when you are close enough to the set point Could you make the simulation itself more realistic For example when the heat pump is heating and all valves are fully open x units of heat are deliv ered to each floor When one of the valves is closed the others still get x units of heat A real system would not work this way closing off one floor would increase the heat flow to the others How could the system be modi fied to better implement what might actually happen Would the control logic have to be altered to accommodate this change This has been a brief introduction to fuzzy systems and the simulation is much simpli fied from the real world However I hope it has piqued your interest in the topic and that you will want to learn more and apply the techniques to your
362. instances and each Thermometer fact would have a reading slot but if the value in that slot changed Jess wouldn t know it The shadow fact s reading slot would never change A more useful Bean includes a mechanism for notifying interested parties that the value of a property has changed The standard JavaBeans mechanism for doing this is via PropertyChangeEvents Jess works best with Beans that send one of these events whenever the value of any property changes It s fairly easy to imple ment this behavior in your own Beans using the helper class PropertyChangeSup port in the java beans package Most of the code you need to write is boilerplate A class that supports PropertyChangeListeners must include the methods addPropertyChangeListener and removePropertyChangeListener they always look the same but they still have to be included in every JavaBean you write Many people create a base class for their Beans that contains a protected Property ChangeSupport member and implementations of these two methods you ll do this here The helper class looks like this package control import java beans public abstract class BeanSupport protected PropertyChangeSupport m_pcs new PropertyChangeSupport this public void addPropertyChangeListener PropertyChangeListener p m_pcs addPropertyChangeListener p public void removePropertyChangeListener PropertyChangeListener p m_pcs removePropertyChangeListener p Writi
363. ion C Jessxx for example The unpacked directory will contain a A Java archive file jess jar which contains the Jess software m The directory examples which contains some simple examples of Jess programs A copy of the Jess manual in the directory docs Open docs index html in your web browser to view the manual To complete your installation add jess jar to your CLASSPATH environment vari able CLASSPATH tells the Java program where to find installed Java software 34 3 1 3 3 1 4 CHAPTER 3 Introducing Jess NOTE The details of setting environment variables vary between platforms For older versions of Windows it involves editing the C AUTOEXEC BAT file For newer Microsoft operating systems you can set environment variables in the System control panel For Unix like systems editing your cshre or profile will do the trick Refer to your operating sys tem s documentation for details CLASSPATH consists of a series of filenames joined by the character your operating system uses to separate filenames in other kinds of lists On Windows this is the semicolon character and on Unix like systems it is the colon Make sure that CLASSPATH at a minimum includes the full path to the jess jar file anda period which represents the current directory So on Windows if there is no pre existing CLASSPATH setting you set this variable to C Jess61 jess jar If you find that your system already
364. ion ident id text text type type 77 and there is no answer for it not MAIN answer ident id j7 and the trigger fact for this question exists ask lt MAIN ask id gt Ask the question bind Panswer ask user text type Assert the answer as a fact assert MAIN answer ident id text answer Remove the trigger retract ask And finally exit this module return I ve explicitly qualified all the fact names with MAIN Although doing so may not be strictly necessary it helps to avoid confusion All of your templates are defined in the module MAIN and therefore they can be shared by all the other modules you define You can test this rule after defining a deffacts to hold a few sample questions You should definitely put this test deffacts into a file rather than just entering it interactively you ll use it again and again to test the rules as you develop them NOTE You should be thinking about putting together a complete test harness now The details here will vary depending on your platform On UNIX you might write a shell script to execute your test code and on a Win dows operating system you might use a BAT file or run the same UNIX scripts using Cygwin The important thing is to make it convenient to 1 Cygwin is a porting layer that lets UNIX tools run on Windows The Cygwin home page is at http www cygwin com 170 10 7 CHAPTER 10 D
365. ion to describe this new XML type and we ll look at some examples I ll also show you an XSLT script that turns the XML version into Jess rules If you haven t seen DTDs before they are a way to indicate the allowed struc ture for a specific kind of XML document DTDs are defined in a kind of pidgin XML a language that is similar to but isn t quite proper XML ELEMENT lines describe individual XML elements and their contents ATTLIST lines list the XML attributes an element is allowed to have A list of rule elements is called a rulebase a rule element has a name attribute and optionally a priority salience attribute A rule s content consists of an 1hs element and an rhs element lt ELEMENT rulebase rule gt lt ELEMENT rule lhs rhs gt lt ATTLIST rule name CDATA REQUIRED priority CDATA gt The means zero or more An lhs element is a list of zero or more pattern and group elements A group is a list of one or more other groups or patterns the means one or more A pattern element is a list of zero or more slot elements patterns and groups have name attributes A pattern can have a binding An rhs is a list of zero or more function call elements lt ELEMENT rhs function call gt lt ELEMENT lhs group pattern gt 368 CHAPTER 20 Jess XML and the enterprise lt ELEMENT group group pattern gt lt
366. ion is long You can t enter a literal long value in Jess as you can in Java Instead you must use the Jess long function to create one from a string In Java you can write long aLongValue 123456789L The equivalent in Jess is bind PaLongValue long 123456789 As you can see even assigning a value to a variable is done using a function call The bind function assigns a value to a Jess variable Let s learn more about vari ables in Jess Variables A Jess variable is a named container that can hold a single value much like a vari able in Java Jess variables however are untyped This means a Jess variable can hold a value of any data type and it can hold values of any number of different types during its lifetime A variable can refer to a single symbol number string list or other value 48 CHAPTER 4 Getting started with the Jess language Most variables in Jess are written as symbols that begin with a question mark The question mark is part of the variable s name Although it is legal to do so it is considered bad style to use any other punctuation marks except dashes or underscores _ in a variable name In particular don t use asterisks because asterisks are used to name global variables which we ll discuss soon You don t need to declare variables in the Jess language they come into being when they are first given a value To assign a value to a variable use the bind function Jess
367. ipping list and so on The recommendations for order query is similar defquery recommendations for order declare variables order recommend order number order part number part product part number part This query lists all the recommendations associated with a given order In this query and the last one part is an internal variable an undeclared variable used only to perform matching within the patterns of the query Maintaining the order number The web application you re going to build needs to give each order a unique number Traditionally this task is handled using a stored procedure in a database A tiny database table holds a single number and a stored procedure increments the value in the table and returns it Using a database makes this process safe for multiple concurrent instances of the web application the database allows only one instance at a time to be updating the order number In this application you ll use the same technique implemented in Jess using a deftemplate with a single slot to represent the tiny table and a deffunction instead of a stored procedure Jess will automatically supply the concurrency control as you ll see in the next chapter Here s the deftemplate deftemplate next order number slot value Here s a query to retrieve the single fact you expect to be using this template defquery order number next order number 17 8 Cleaning up 303 And finally h
368. is an index for example you would pass the value 3 to getTemperature to get the temperature reading from sensor number 3 By convention all these index numbers start at 1 not 0 Native methods Of course you re writing the software in Java so you need to make the connec tion between your Java code and the C library This is easy enough to do using the Java Native Interface JNI A native method is a method of a Java class that happens to be written in some system level language like C Native methods use a special JNI library to bridge the gap between the Java virtual machine and the real machine outside The knowledge that your program will be dealing with external hardware can be encapsulated inside a single Java class that class will have one native method for each of the C library functions Writing native methods is outside the scope of this book The best reference for writing them is probably Sun s web based Java tutorial http java sun com docs books tutorial native1 1 index html For our purposes all you need to know is that from the Java side you declare a native method like this public native void myMethod String myArgument 14 2 2 Defining the hardware interface 225 A native method has no body just like an abstract method and includes the key word native in its declaration otherwise there s nothing special about it Java expects to find the definition of this method in an external library you must provi
369. is chapter you ll modify the ask module yet again to connect it to your GUI In particular you need to modify the function ask user Listing 13 1 is a new version of ask user that displays the questions in the JTextArea instead of at the command line If you run the whole application with this modi fied ask user you ll need to look at the JFrame to see the questions and at the command line to type your answers You ll return to this function soon the final version will no longer read from the command line Diagnostic Assistant lolx lease wait Figure 13 2 Application window with an added JTextArea for displaying questions Getting answers 207 Listing 13 1 ask user function to display questions changed lines in bold deffunction ask user question type valid Ask a question and return the answer bind answer while not is of type Panswer type valid do bind s question if eq type multi then bind s str cat s crlf Valid answers are foreach Pitem valid bind s str cat s item bind s str cat s qfield setText s bind answer read return answer 13 4 Getting answers Now that your application displays questions to the user via its new GUI it ought to collect answers the same way Currently the input routine ask user reads answers as strings the user types at the command line As a first step you ll modify the ask module to use a
370. is chapter you ll Design deftemplates for the Tax Forms Advisor a Partition the application with defmodules a Write code to ask questions of the user 161 162 10 1 CHAPTER 10 Designing the application In this chapter you will begin to develop the Tax Forms Advisor system described in chapter 9 You will decide what the facts should look like and how to divide the rules into modules when you write them in the next chapter You ll also design some I O functions and other infrastructure the rules in the system need In chapter 11 you ll write the actual rules on the foundation you develop here The design process you ll follow in this chapter is idealized There are no false starts or backtracking In truth designing a system like this usually involves exper imentation especially when you re still gaining experience Don t be discouraged on the contrary feel free to experiment with different approaches to implement ing this application and to the others in this book In previous chapters of this book you ve entered code directly at the Jess gt prompt This approach is great for experimenting but when you re developing an application you ll want to save the code in a text file instead You can then execute your code either by using Jess s batch function which executes the contents of a file full of Jess code or by specifying the filename on the command line like this C gt java classpath jess jar jess
371. is example the keys are the names of grocery items and the values are the prices Jess gt call prices put bread 0 99 Jess gt call prices put peas 1 99 Jess gt call prices put beans 1 79 Jess gt call prices get peas 1 99 The first argument to call is a Java object and the second argument is the name of a method to invoke The remaining arguments to call are the arguments to be passed to the Java method The arguments are converted to Java types according to table 5 1 In this example you ignore the return value of HashMap get and allow Jess to simply display it Often though you ll want to do something with the return type binding it to a variable or calling another method on it in turn are two common alternatives Jess converts the return values of Java methods to Jess types according to table 5 2 These conversions are generally the reverse of those in table 5 1 5 2 1 Calling Java methods 65 Table 5 2 Standard conversions from Java types to Jess types Java type Jess type A null reference The symbol nil A void return value The symbol nil String RU STRING boolean or java lang Boolean The symbol TRUE or FALSE byte short int or their wrappers RU INTEGER long or java lang Long RU LONG double float or their wrappers RU FLOAT char or java lang Character RU ATOM a symbol An array A list Anything else RU EXTERNAL_ADDRESS You can call virtually any
372. is ok This rule corresponds to the END oval car starts yes gt assert car is ok The program you are building will diagnose problems with computers so you should be able to represent the knowledge you gather in a series of flowcharts If you were gathering the knowledge by interviewing a computer repair technician you could start by asking her to list some of the most common computer prob lems she deals with For each of these problems you d develop a flowchart around the hypothesis that this particular problem was affecting the broken com puter Each of these flowcharts would show the series of tests the technician would perform to attempt to verify or disprove the hypothesis It should be easy to imagine how the interview would be conducted For each flowchart you d ask What would you do first to diagnose this problem and what are the possible outcomes The most likely follow up questions would then be of the form What would you do if you saw outcome X when you did Y In between of course you d need to ask for clarification about what various technical terms meant and these answers would be collected in prose form Because each step in the flowchart might be fairly complex you d probably want to give each box an identifier like F10 3 meaning the third numbered box on flowchart number 10 and then record the contents of each box in prose form on another piece of paper A real diagnostic system mi
373. is reexecuted If the property is set to FALSE or nil then reset does not affect global variables set salience evaluation when defined when activated every cycle Changes the current setting of the salience evaluation behavior By default a rule s salience is determined once when the rule is defined when defined If this property is set to when activated then the salience of each rule is redeter mined immediately before each time it is placed on the agenda If the property is set to every cycle then the salience of every rule is redetermined immediately after any rule fires this can be very computationally expensive set strategy depth breadth lt lexeme gt Lets you specify the con flict resolution strategy Jess uses to order the firing of rules of equal salience Two strategies are built into Jess depth LIFO and breadth FIFO When the depth strategy is in effect the default more recently activated rules are fired before less recently activated rules of the same salience When the breadth strategy is active rules of the same salience fire in the order in which they are activated Note that in either case if several rules are activated simultaneously by the same fact assertion event the order in which they fire is unspecified implementation dependent and subject to change 410 APPENDIX A Jess functions You can implement your own strategies in Java by creating a class that imple ments the jess Stra
374. ist directly at the Jess prompt Jess assumes it s a func tion call if you nest a list inside a function call Jess assumes that nested list is a 50 4 3 CHAPTER 4 Getting started with the Jess language function call too So how can you create a list that s nota function call You use spe cial functions like create to make a list and then you bind the list to a variable Jess gt bind grocery list create eggs bread milk eggs bread milk The variable grocery list now holds a list of three items This kind of list which just contains data is called a plain list Plain lists are use ful data structures in Jess They re a lot like Java arrays You can access the ele ments of a Java array using square brackets with a numeric subscript For example arr 0 is the first element of an array named arr To access an element of a plain list in Jess you call the nth function which returns a single list element Jess gt printout t nth 2 grocery list crlf bread You may have noticed that many of the Jess functions that deal explicitly with lists have names ending in this is just a convention but a useful one Two other list related functions are first and rests Jess gt first grocery list eggs Jess gt rest grocery list bread milk The function first returns a list containing just the first element of its single argument and rest returns a list holding the second and subsequent elements Y
375. it is explicitly defined with a deftemplate or defclass construct then you must define it before calling do backward chaining You can use do backward chaining on ordered deftemplates before they are created however Once you have declared your reactive deftemplates you can define rules with patterns that match facts of the corresponding types Note that you must call do backward chaining before defining any rules that use the template This rule prints the factorial of 10 assuming a fact recording this information exists Jess gt defrule print factorial 10 factorial 10 r1 gt printout t The factorial of 10 is r1 crlf TRUE Patterns that match backward chaining reactive deftemplates are called goals When the rule compiler sees a goal pattern it rewrites the rule and inserts some special code into the internal representation of the rule s LHS If when the rule engine is reset there are no matches for this pattern the code asserts a fact into working memory that looks like this need factorial 10 nil The head of the fact is constructed by taking the head of the reactive pattern and adding the prefix need These need x facts are called goal seeking or trigger facts This particular trigger fact means that another fact factorial 10 is needed to Backward chaining rules 119 satisfy some rule Jess got the number 10 directly from the pattern in print factorial 10 nil is a placeholder that means any value
376. ite your own control structures Here s an implementation of a Userfunction named n times that evaluates its second argument a number of times specified by its first argument import jess public class NTimes implements Userfunction public String getName return n times public Value call ValueVector vv Context c throws JessException int count vv get 1 intValue c Value expr vv get 2 for int i 0 i lt count i expr resolveValue c return Funcall TRUE You can compile this class load the function into Jess and call it Jess gt load function NTimes TRUE Jess gt n times 10 printout t kk kK KK ATRUE Jess gt Each time the Java method resolveValue is called the Jess function in expr is evaluated and so this line of code prints 10 asterisks This is how Jess s built in control structures like while and foreach are implemented Returning a value 245 15 3 Returning a value 15 3 1 The call method returns a Value object You can return one of call s argu ments return one of a handful of constants in the jess Funcall class or create a new Value of your own You should never return one of cal1 s arguments directly rather evaluate the argument by calling resolveValue on it and return the result instead Failing to do this will result in undefined behavior The Funca11 class contains a few constants you can use as function return values a TRUE Boolean true
377. ithout pro viding extra information in the form of a definstance call the RI has to make some assumptions and provide the extra information Every time an object is added to working memory Jess automatically defines a defclass for it the def class name is the full class name of the object java Lang Integer for example Therefore a program to add two Integers would look like this defclass java lang Integer java lang Integer nil defrule add two integers java lang Integer OBJECT i1 java lang Integer OBJECT i2 amp i1 gt undefinstance i1 undefinstance i2 bind i3 new Integer il intValue i2 intValue definstance java lang Integer i3 static Working memory elements that are not JavaBeans call for a different program ming style compared to normal Jess You can use individual patterns to bind objects as is the case here but most tests will be fairly verbose and so should be expressed as test conditional elements as in the following example defrule find number 3 java lang Integer OBJECT i1 test eq 3 i1 intValue gt Found 3 do something One other thing about the RI is special The input files must be XML rather than normal Jess input files The format is exceedingly simple Each XML file must contain a rule execution set element which contains three children a name the name of the set of rules a description a brief summary of the rule s pur po
378. iting your own Jess commands you may want to create your own Jess Exception objects to report errors The good news is that Jess automatically takes care of setting the line number program text and Jess stack trace information after your JessException is thrown all you need to worry about are the routine name error message and possibly a nested exception JessException has three constructors which differ only in their last argu ment The first two arguments are always a routine name and a message The rou tine name indicates where the exception was created It can be either the name of a Java class and method or the name of a Jess language function use your best judgment The routine name you use is displayed on the first line of the message returned by JessException toString The second argument to each JessException constructor is the error mes sage a short description of what went wrong perhaps five words at most For two of the three constructors the third argument is simply more data to be appended to the message with a space in between One of the constructors accepts an int which is handy for reporting number related errors and the other accepts a sec ond string For the third constructor the last argument is a nested exception This code snippet is from an imaginary Userfunction that implements a goo gle function in Jess It tries to connect to the Google web site and reports any fail ure via a JessException try S
379. ition any argu ment or return value types must either be serializable or be themselves remote interfaces The remote interface for the factory therefore looks like this import java rmi interface JessFactory extends Remote Jess create throws RemoteException Jess is the name you ll give the remote Rete interface You can make this inter face as simple or complex as desired A version that had only a single execute Command method would be quite useful To make it more general however it at least needs a way to transfer Java objects back and forth because executeCommand only accepts strings as an argument To do this you can add support for calling store and fetch Here s the resulting interface import java rmi import jess interface Jess extends Remote Value executeCommand String command throws RemoteException JessException void store String name Object object throws RemoteException Value fetch String name throws RemoteException You might want to add more methods but this simple interface gives you access to almost all of Jess s capabilities Implementing the interfaces Implementing these two interfaces is easy because all they do is call a constructor JessFactory or forward methods to a Rete object Jess Here s an implemen tation of the JessFactory interface import jess import java rmi import java rmi server An RMI based rule server 381 public class Jess
380. itten as a declarative program A rule based pro gram doesn t consist of one long sequence of instructions instead it is made up of discrete rules each of which applies to some subset of the problem A few rules plucked from the robot s driving program might look like these IF the engine has stalled THEN start car END IF you hear sirens AND you are driving THEN pull over to curb END IF you see brake lights AND you are getting closer to them THEN depress the brake pedal END In a rule based program you write only the individual rules Another program the rule engine determines which rules apply at any given time and executes them as appropriate As a result a rule based version of a complex program can be shorter and easier to understand than a procedural version Writing the pro gram is simpler because you can concentrate on the rules for one situation at a time Modifying the program is also simpler if you ve ever had to work on a Rules and rule engines 17 program containing a dozen levels of nested if statements you ll understand why In the rest of this chapter we ll formalize some of the ideas behind rule based systems and see how they are constructed 2 2 Rules and rule engines A rule is a kind of instruction or command that applies in certain situations No chewing gum in school no running with scissors and other rules of that ilk are some of the first explicit rules
381. iving with the dependencies function The dependents function tells you what facts are depen dent on another given fact Both functions take either a single fact object or an integer fact ID as an argument If one or more logical CEs appear in a rule they must be the first patterns in that rule a logical CE cannot be preceded in a rule by any other kind of CE You can use the logical CE together with all the other CEs including not and exists A fact can thus be logically dependent on the nonexistence of another fact or on the existence of some category of facts in general Shadow facts from definstances are no different than other facts with regard to the logical CE Shadow facts can provide logical support and can receive logi cal support Backward chaining rules The rules you ve seen so far have been forward chaining rules as I ve said that means the rules are treated as if then statements with the engine simply exe cuting the RHSs of activated rules Some rule based systems notably Prolog and its derivatives support backward chaining In a backward chaining system rules are still if then statements but the engine actively tries to make rules fire If the if clause of one rule is only partially matched and the engine can determine that fir ing some other rule would cause it to be fully matched the engine tries to fire that second rule This behavior is often called goal seeking Backward chaining rules 117 As an
382. k Agents were running on each machine and they used a sort of post and bid method to decide which machines would run which piece of a computation based on machine capabilities and load balancing The agents were fairly intelli gent in their decision making capabilities and the plans they developed were sometimes surprising Their brains were rule engines software systems that used rules to derive conclusions from premises That project led to others and soon I developed an interest in mobile agents software entities that can travel from node to node in a computer network main taining their state as they go Thus was born the idea for a rule engine whose state could be packaged up sent across a wire and reconstituted The newly released Java language seemed to be a perfect vehicle for this rule engine and such was the origin of Jess the rule engine for the Java Platform Jess is a general purpose rule engine developed at Sandia National Laborato ries Written in the Java programming language Jess offers easy integration with 1 Jess is a registered trademark of the Sandia Corporation xxi xxii PREFACE other Java based software Jess is free for academic and government use and it can be licensed for commercial use You can download a fully functional Home Edition of Jess free of charge if you own a copy of this book see chapter 3 for download instructions You can use the Jess Home Edition for noncommerc
383. kets are also used to group expressions together so that one of the repeating operators can be applied to a group The basics Just as books are made up of words code is made up of tokens A token is a sequence of characters that a computer language recognizes as a fundamental unit Jess understands only a few different kinds of tokens symbols numbers strings and comments Once you know about these you can begin to do some programming Whitespace Input to Jess is free form Newlines and whitespace are not significant except inside quoted strings You can use newlines and indentation to highlight the struc ture of the code you write just as in Java for example the following code if lt x 3 then printout t x is less than three crlf would usually be written if lt x 3 then printout t x is less than three crlf Symbols The symbol is a core concept of the Jess language Symbols are tokens that are very much like identifiers in Java A Jess symbol can contain letters numbers and the following punctuation marks lt gt _ and A symbol may 4 1 3 4 1 4 The basics 43 not begin with a number Some of the punctuation marks and cannot be used as the first character of a symbol but can appear in other positions The other punctuation marks listed can appear as the first character or any character of a symbol Jess symbols are case sensitive foo FOO and Foo are all di
384. kets to use in each individual hash table A small value saves memory possibly at the expense of performance a larger value uses more memory but leads up to a point to faster pattern matching The default used if you don t declare an explicit value is cho sen for optimal performance In general you might want to declare a large value for a rule that is likely to gen erate many partial matches Conventional wisdom has it that prime numbers are the best choices Experimentation is the only way to determine the best value for a particular rule If nis the number of item facts then the following rule will need n tokens in the left memory of its third join node it s an excellent candidate for a large node index hash value if you expected there to be more than a few item facts 142 8 6 8 6 1 CHAPTER 8 Under the hood how Jess works Jess gt defrule many matches declare node index hash 167 item c printout t a pb c d crlf The set node index hash function sets the default value for this parameter used for rules without a specific declaration If you don t set a value the default is 101 More complexity and initial fact A number of pattern matching situations can t be handled by what we ve discussed so far Studying them is enlightening because it explains some otherwise seemingly arbitrary properties of software like Jess that is based on the Rete algorithm Implementing the no
385. l but you should keep an eye out for the problem during testing The vent rules Now you re ready to write the vent rules The following rule says that if the heat pump for a floor is in cooling mode the vent is closed and the temperature is above the set point then the vent should be opened defrule floor too hot vent closed HeatPump state cooling number p 16 5 Testing the whole system 263 Vent state closed floor f amp eq p which pump f Thermometer floor f reading t amp gt t set point gt set vent state f open Similarly if the heat pump for a floor is cooling the floor is too cold and the vent is open then the vent should be closed defrule floor too cold vent open HeatPump state cooling number p Vent state open floor amp eq p which pump f Thermometer floor f reading t amp lt t set point gt set vent state f closed You need analogous rules floor too cold vent closed and floor too hot vent open dictating what to do when the heat pump is in heating mode these aren t shown here but they re very similar to the other pair The four rules corre spond to the four cells in table 16 3 that have an entry other than Do Nothing Testing the whole system The first version of the HVAC Controller system is now essentially complete All that remains is to clean things up a bit and test it in operation You can
386. l still be hotter than the rest and will need cooling instead Make sure the floors that need heating stay between 68 and 70 degrees while the upper floors stay between 70 and 72 Next try an outdoor temperature of 20 degrees The upper floors will lose enough heat to the outside that they too will require heating Finally try outdoor temperatures very close to the set point Try several tempera tures from 67 to 73 watching carefully for illegal transitions from heating to cooling You can experiment with different numbers of floors by specifying larger or smaller numbers in the call to init simulator the number of floors should be divisible by 3 You should find that the HVAC Controller system has no problem handling a building with 99 floors or more Remember that 99 floors corresponds to 99 Thermometers 99 Vents and 33 HeatPumps for a total of 231 constantly changing inputs a reasonably complex system Each Bean runs in its own Java thread demonstrating Jess s ability to run well in a multithreaded environment with 99 floors Jess is receiving asynchronous property change events from 231 dif ferent threads The system you ve built works fairly well although because the vents can only be completely open or completely closed the temperature at each floor fluctuates by as much as one or two degrees If the vents were continuously variable the temperature could be controlled more closely In the next section we ll look at one
387. lass server This makes it easier to modify the server over time because otherwise every client installation would need to be updated with a new set of stub files every time the server changed To enable this mechanism you need to install a special RMI security manager define a security policy and set sev eral system properties defining the location of the class server You can read all about this and other details in Sun s RMI tutorial http java sun com docs books tutorial rmi index html JSR 94 the javax rules API The javax rules package is a standard lightweight API for attaching to and using a rule engine It can be used in both J2SE and J2EE applications It includes mechanisms for storing and managing rule bases as well as classes that represent individual sessions with a rule engine Using the javax rules package you can write Java code to work with rule engines in a generic manner so that changing from one rule engine to another won t necessitate changes to the code The name JSR 94 refers to the standards committee that developed this new API its mem bers represent many of the major rule engine vendors I was a member of the committee javax rules is nota standard rule language it says nothing about rule lan guages Therefore if you change rule engine vendors you will probably have to 384 21 3 1 CHAPTER 21 Jess in the J2EE environment rewrite your rules even if you don t need to change your Java code
388. ld try it out as a user and offer advice to correct its performance If possible a prototype of the system should be presented to the expert at every interview except perhaps the first one Likewise development needn t be deferred until knowledge engineering is complete For many small projects the knowledge engineer is one of the develop ers and in this case you may be able to dispense with the cards and simply encode the knowledge you collect directly into a prototype system This is what you ll do for the Tax Forms Advisor More on writing cards To write down the rule use ez form on a card I had to make up the deftemplate names filing status and income and also define an imaginary function recommend In general you will write rules on these cards in pseudocode they re meant to suggest how the real rules might be coded but they re just guides When actual development begins on the system these early guesses will help the devel opers figure out what deftemplates and other infrastructure they need to define Finally note that although I wrote use ez form in Jess syntax it would be per fectly OK for a knowledge engineer to use natural language or pseudocode that looks like some other programming language If you are a knowledge engineer but not a programmer writing rules in your native language may be the only option and that s fine Interviews People are the best source of information about the requirements for a system
389. le choice questions Next I developed and tested the trigger module trigger can and should be tested first in isolation that is without ask loaded in to complicate matters To test trigger I wrote one dummy rule with an initial fact pattern and one answer pattern Then I used watch a11 to confirm that executing the reset command made the trigger module come into focus the supply answers rule fire and the appropriate MAIN ask fact appear in working memory Then as I translated each new flowchart into rules I ran the application and answered questions based on a set of planned diagnostic scenarios confirming that the diagnosis was correct in every case As discussed in section 11 2 3 don t be discouraged by the thought that you should have 100 test coverage Remem ber that some testing is infinitely better than none at all One issue that came up during this kind of testing was that the hardware question which you want to be asked first isn t asked until the end of the pro gram Paradoxically this happens because you define the right architecture question fact first The need answer trigger fact for this rule is generated as soon as the rule is defined and under Jess s default depth conflict resolution strategy later activations fire before earlier ones Thus although the decision to ask the hardware question is made first it is overridden by the need to ask other 202 12 7 CHAPTER 12 Writing the PC Repair Assistan
390. le m_ventFV null static try m_ventFV new FuzzyVariable vent 0 1 catch InvalidFuzzyVariableNameException nameEx NOTHING catch InvalidUODRangeException rangeEx NOTHING try m_ventFV addTerm fullyClosed new SingletonFuzzySet 0 0 m_ventFV addTerm slightlyOpen new SingletonFuzzySet 0 25 Vv Controlling with fuzzy rules 273 m_ventFV addTerm halfOpen new SingletonFuzzySet 0 5 A m_ventFV addTerm mostlyOpen new SingletonFuzzySet 0 75 m_ventFV addTerm fullyOpen new SingletonFuzzySet 1 0 catch XValuesOutOfOrderException outOfOrderEx NOTHING catch XValueOutsideUODException outsideOUD NOTHING public Vent Hardware hw int floor m_hardware hw m_floor floor public static FuzzyVariable getVentFuzzyVariable return m_ventFV public int getFloor return m_floor public double getState Modify method to return m_hardware getVentState m_floor return numeric value public void setState double state m_hardware setVentState m_floor state This part of the code creates a new FuzzyVariable to represent the vent position and loads the linguistic terms into the variable Note that you no longer define the Vent Bean to implement Runnable because you do not need to provide Jess with updated values of the vent state manual vent adjustments will no longer be allowed Some oth
391. le to a Create a simulator including a GUI Get the number of floors and heat pumps Determine which heat pump goes with which floor Operate the heat pumps and vents Let s tackle these tasks one at a time in this order Creating a simulator The first command you will write creates a hardware simulator attaches a GUI to it and stores it somewhere where the other functions can access it If you assume that each individual instance of Jess each individual jess Rete object controls only one building s systems at a time then storing a reference to the simulator in the Userfunction object itself will work well Name the Jess function init simulator here is the implementation package control import jess class InitSimulator implements Userfunction 15 5 2 The HVAC functions 249 static final String NAME init simulator private Simulator m_simulator public static Simulator getSimulator Context c Rete engine c getEngine InitSimulator is InitSimulator engine findUserfunction NAME return is m_simulator public String getName return NAME public Value call ValueVector vv Context c throws JessException int nFloors vv get 1 intValue c m_simulator new Simulator nFloors new SimulatorGUI m_simulator return new Value m_simulator This function is fairly straightforward It accepts one argument the number of floors in the simulated building The Userfunc
392. led heuristics for medical diagnosis engi neering chemistry and computer sales One of the early expert system successes was MYCIN a program for diagnosing bacterial infections of the blood Expert systems had a number of perceived advantages over human experts For instance unlike people they could perform at peak efficiency 24 hours a day forever There are numerous dramatic examples in the computer science litera ture of these early systems matching or exceeding the performance of their human counterparts in specific limited situations Predictions were made that someday sophisticated expert systems would be able to reproduce general human intelligence and problem solving abilities Over time of course the drama receded and it became clear that researchers had vastly underestimated the complexity of the common sense knowledge that underpins general human reasoning Nevertheless excellent applications for expert systems remain to this day Modern expert systems advise salespeople sci entists medical technicians engineers and financiers among others Today general rule based systems both those intended to replace human expertise and those intended to automate or codify business practices or other activities are a part of virtually every enterprise These systems are routinely used to order supplies monitor industrial processes prescreen r sum s route telephone calls and process web forms Many commercial application ser
393. les about sound 197 Degrading gracefully 198 To boot or not to boot 199 RAM problems 199 Questioning authority 200 12 6 Testing 201 12 7 Summary 202 CONTENTS Xv Adding a graphical interface 203 13 1 13 2 13 3 13 4 13 5 13 6 13 7 13 8 Getting started 204 Displaying a window 205 Displaying questions 206 Getting answers 207 The main thread vs the event thread 209 Better input components 2 3 Finishing touches 214 Testing the interface 215 Summary 216 The reality connection 221 14 1 14 2 14 3 14 4 14 5 The system 223 Defining the hardware interface 224 Native methods 224 Writing a simulator 225 Simulating getTemperature 227 Adding a graphical interface 229 Writing the JavaBeans 229 Rules about Thermometers 232 Writing the other Beans 233 JavaBeans and serialization 235 Summary 235 Extending the Jess language 237 15 1 15 2 The Userfunction interface 238 The getName method 239 The call method 239 Loading a Userfunction into Jess 239 Handling arguments 240 How many arguments 240 Using arguments 241 Resolving variable arguments 242 Resolving function call arguments 243 XVi CONTENTS 15 3 15 4 15 5 15 6 15 7 Returning a value 245 Constructing Value objects 245 Beyond simple examples 246 Holding state 246 Multiple personalities 246 Userfunctions and serialization 247 Grouping functions with Userpackage 247 The HVAC functions 248 Creating
394. letContext servletContext getServletContext Rete engine Rete servletContext getAttribute engine String factsFileName servletContext getInitParameter factsfile File factsFile new File factsFileName File tmpFile File createTempFile facts tmp factsFile getParentFile 19 14 Deploying the application 351 engine executeCommand save facts tmpFile getAbsolutePath order recommend line item next order number factsFile delete tmpFile renameTo factsFile catch Exception je Log error This implementation tries to be careful about losing data it writes the facts to a tem porary file and then renames the temporary file to factsfile only if it succeeds The method checkInitialized is called by all the servlets but you want this version of destroy to be called only once I put it into the Catalog servlet although it could go into any of them After extensive use writing and reading the accumulated facts from flat files could become prohibitively slow Also loading all the data at startup and saving it all at shutdown may incur too great a risk of data loss in the event of a crash Instead of using a flat file the application could be changed to use a relational database Orders relevant to a particular user could be loaded from a database on demand by the Recommend servlet The Purchase servlet could save those facts rel evant to a particular order just as the order
395. ling a rule query or deffacts definition it looks for tem plates in three places in order 1 Ifa pattern explicitly names a module only that module is searched 2 Ifthe pattern does not specify a module then the module in which the rule is defined is searched first 3 If the template is not found in the rule s module the module MAIN is searched last Note that this makes the MAIN module a sort of global namespace for templates The example in listing 7 5 illustrates each of these possibilities In this example three deftemplates are defined in three different modules MAIN mortgage payment WORK job and HOME hobby Jess finds the WORK job template because the rule is defined in the WORK module It finds the HOME hobby template because it is explicitly qualified with the module name And the MAIN mortgage payment template is found because the MAIN module is always searched as a last resort if no module name is specified Jess gt clear TRUE Jess gt assert MAIN mortgage payment 2000 lt Fact 0 gt Jess gt defmodule WORK TRUE Jess gt deftemplate job slot salary TRUE Jess gt defmodule HOME TRUE Jess gt deftemplate hobby slot name slot income TRUE Jess gt defrule WORK quit job job salary s HOME hobby income i amp gt i s 2 mortgage payment m amp lt m i 126 7 6 3 CHAPTER 7 Writing rules in Jess Jess gt ppde
396. ll NIL STRING d addSlot address Funcall NIL STRING engine addDeftemplate d String data Joe Smith 123 Main Street Fred Jones 333 Elm Circle Bob Weasley 211 Planet Way for int i 0 ix lt data length i Fact f new Fact person engine f setSlotValue name new Value data i 0 RU STRING f setSlotValue address new Value data i 1 RU STRING engine assertFact f engine executeCommand facts In listing 18 1 you specify a value for every slot in each Fact but doing so is not required If you don t specify a value for a slot the default value is used when the Fact is asserted Once you assert a Fact object you no longer own it it becomes part of the Rete object s internal data structures As such you must not change the values of any of the Fact s slots If you retract the fact the Fact object is released and you 18 2 1 18 2 2 Working with Fact objects in Java 313 are free to alter it as you wish You can retract your fact from the engine using the retract method Multislots Jess facts can contain multislots single slots that hold multiple data items You can add a multislot to a Deftemplate using the addMultiSlot method You can then set the value of that slot in a fact as usual but the slot value must be a jess Value of type RU LIST A LIST value contains a ValueVector the elements of the val ueVector are th
397. ll from a string m foreach Executes a block of code once for each item in a list m if then else Chooses among alternative courses of action u progn Executes a group of function calls m while Executes a block of code while an expression is true You ll use some of these control flow functions like foreach while and if then else all the time and some of them only rarely We ll look at the most common ones first foreach The foreach function provides the perfect way to shop for each item on your gro cery list It evaluates a block of expressions once for each element in a plain list Each time through the block a variable you supply is set equal to the correspond ing entry from the list you also supply The syntax diagram looks like this foreach lt variable gt lt list gt lt expression gt t Here s an example of using foreach to print each item on your grocery list on a separate line Jess gt bind grocery list create eggs milk bread eggs milk bread Jess gt foreach e grocery list printout t e crlf eggs milk bread The foreach function is useful when you already have a list of values for the loop variable Often though you need a more general kind of looping construct such as while 52 4 3 2 4 3 3 CHAPTER 4 Getting started with the Jess language while The while function lets you repeatedly evaluate a group of one or more expres sions as long as some condition remains true
398. lls printout to print to standard output and the argument cr1f starts a new line like printing a n in Java This parenthesized list of symbols is the Jess way of calling a function Jess immediately evaluates the call to printout producing the side effect of printing Hello World to the console In addition to jess Main the class jess Console presents the same com mand line interface in a graphical console window see figure 3 1 If you would like a more elaborate graphical interface you can try JessWin a free graphical development environment for Jess written by William Wheeler JessWin which is also written in Java is menu driven and contains an integrated graphical edi tor see figure 3 2 You can download JessWin from the Jess web site http herzberg ca sandia gov jess user html Jess Console Jess the Java Expert System Shell Copyright C 2001 EJ Friedman Hill and the Sandia Corporation Jess Version 6 1b3 3 1 2003 Jess gt printout t Hello World cri Hello World Jess gt Clear Window Figure 3 1 The jess Console interface for Jess This as an alternative to the command line jess Main interface You can enter Jess code in the small text field at the bottom left and Jess output appears in the scrolling text area at the top 36 CHAPTER 3 Introducing Jess JessVin Watch Options Help Start Reset cri u pert System Shell mane cue 1 E J Friedman Hill and the Sandia Corporation S 3 3 1 20
399. lready contains all the acceptable x y pairs so the correct x y z triples can be formed without repeat ing the pattern matching computation done on the first cycle One new activation will be created for each precomputed x y pair You can now see how the Rete architecture lets you avoid repeating computation over time 8 3 2 8 4 Easy optimizations for Rete 139 Handling retract So far you ve seen how the Rete algorithm can be used to efficiently handle the pattern matching that happens during assert commands but what about retract Rete can handle removing activation records as easily as it can handle creating them The trick to doing so is that you don t send facts through the net work You send tokens A token is an association between one or more facts and a tag or command The tag tells the individual nodes how to interpret the token Jess uses four different tags defined as constants in the jess RU class ADD REMOVE CLEAR and UPDATE ADD is used for asserting facts as you ve already seen The behavior described so far only applies for token with a tag value of ADD The REMOVE tag is used for retractions If a REMOVE token arrives at a join node the node looks in the appropriate memory to find a matching token If it finds one the token is deleted All allowed pairings between that token and all the tokens in the opposite memory are then composed also with the REMOVE tag These tokens are sent
400. lt html gt This servlet does nothing but display Hello World on an HTML page but obvi ously the page could be far more complex it could also contain data computed on the fly so that the page was different each time Servlets can contain other methods and other classes so the sky s the limit as far as the content of the gener ated page goes Beside merely displaying a page servlets can receive query data from the user and react to it They can also load configuration data and base their behavior on that They can be invoked by other servlets and invoke still other servlets in turn I ll describe how all this works as it comes up in this chapter The HttpServletRequest object provides input to the servlet like the afore mentioned query data usually from HTML forms The HttpServletResponse object represents the page that s being generated This example only uses the setContentType and getWriter methods The method setContentType speci fies the MIME type of the data you re sending back from the servlet often text 326 CHAPTER 19 Deploying web based applications html to indicate an HTML document The method getWriter returns a Print Writer to which you can send the text that makes up the HTML page 19 2 J2EE and the Tomcat engine 19 2 1 Now that you have a simple servlet how can you set things up so it can be accessed on the Web You need to compile it and then deploy it into a servlet container A
401. lved automatically by methods like intValue But now you should see why using resolveValue is impor tant not using it could result in function call arguments being invoked multiple times 244 CHAPTER 15 Extending the Jess language A similar but subtler problem arises if you try to store an argument in a Java data structure or otherwise arrange to have a Value object live longer than the body of your call method If you do this with a value of a simple type there are no problems But if you store a value of type RU VARIABLE or RU FUNCALL it often won t be possible to resolve the value later because you can t resolve a variable without the relevant jess Context object If you look back at the implementation of bind in the previous section you ll see that before using the second argument as an argument to setVariable I was careful to call resolveValue to get the real value of the argument That way calling bind x y 1 sets x equal to one greater than the current value of y If bind didn t use resolveValue x would hold the function call y 1 Accessing x might give a value that changed over time as y changed or it might give an error if y isn t defined when x is read Just as bind s implementation takes advantage of Jess s lazy resolution to allow it to set the value of a variable control structures use it to allow them to calla function multiple times You can take advantage of this functionality to wr
402. lways used to connect the first pattern on a rule s LHS to the network Passing the mouse over a node dis plays information about the node and the tests it contains Double clicking on a node brings up a dialog box containing additional information for join nodes the memory contents are also displayed the same information displayed by the matches function and for terminal nodes a pretty print representation of the rule is shown You can move the individual nodes around by dragging them with your mouse In figure 8 4 the nodes have been dragged into position by hand to resemble the diagrams in the other figures Exploring the Rete network in Jess 145 js Network View ASA Figure 8 4 Home Jess s view command displays i the Rete network in a window To fully appreciate what s happening in the view command s display you need to know something about how Jess implements the Rete algorithm Jess s network is literally a network of interconnected Node objects Network nodes are repre sented by subclasses of the abstract type jess Node Pattern network nodes which perform tests on single facts have Node1 as part of their name There are also three classes of join nodes jess Node2 jess NodeNot2 and NodeTest that implement normal joins the not conditional element and the test condi tional element respectively Terminal nodes are instances of jess NodeTerm There are quite a few different Nodel classes jess NodelTECT
403. m the doGet method of each servlet you write To make it easy to call you can put it in a base class BaseServlet and let all your serv lets extend BaseServlet You ll be able to put other common behavior in BaseServlet as well checkInitialized looks like this protected void checkInitialized throws ServletException ServletContext servletContext getServletContext String rulesFile servletContext getInitParameter rulesfile 338 19 8 2 CHAPTER 19 Deploying web based applications String factsFile servletContext getInitParameter factsfile if servletContext getAttribute engine null try Rete engine new Rete this new Batch batch rulesFile engine engine reset if new File factsFile exists engine executeCommand load facts factsFile servletContext setAttribute engine engine catch Exception je throw new ServletException je This listing introduces the ServletContext class Every web application has a sin gle ServletContext shared by all the servlets in the application get Init Parame ter returns named configuration values from the web xml file here they re used to provide filenames for the rules and the catalog data The methods getat tribute and setAttribute let servlets store arbitrary Java objects in the Servlet Context these objects last as long as the web server is running The checkInitialized method creates a Rete object loads it
404. mber player order customer id id order number currentOrder amp gt currentOrder orderl1 not order customer id id order number order3 amp gt order3 currentOrder The last two patterns narrow the matches a bit The first pattern eliminates video tapes or DVDs the customer has purchased or is currently purchasing and the final pattern ensures that you only recommend one videotape or one DVD per order not line item customer id id part number media not recommend order number currentOrder type sym cat discretionary c In the previous section all the recommendations were of type requirement The recommendations you re creating here are discretionary videotape and dis cretionary dvd disk Other parts of the system will therefore be able to tell the difference between true requirements and mere sales pitches You compose the correct category name using the sym cat function Finally if all the patterns match you ve identified a videotape you can recom mend to the customer gt order number currentOrder because name part number media type sym cat discretionary c assert recommend This rule simply asserts a recommend fact as recommend requirements did and leaves it up to other software to communicate with the user Conspicuous consumption If one DVD or videotape is good more must be even better The rule recommend more media asks t
405. me a new ver sion of Jess is released Review of the problem Before looking at Jess s implementation let s review the problem Jess is meant to solve Jess is a shell for rule based systems In the simplest terms this means Jess s purpose is to continuously apply a set of if then statements rules to a set of data the working memory Each system built from the shell defines its own rules Jess rules look like this defrule library rule 1 book name x status late borrower y borrower name y address z send late notice x y z This rule might be translated into pseudo English as follows Library rule 1 IF A late book X exists borrowed by a person Y AND That borrower s address is known to be Z THEN Send a late notice to Y at Z about the book X END The information about books and borrowers would be found in the working memory which is therefore a kind of database of bits of factual knowledge about An inefficient solution 135 the world Entities like books and borrowers are called facts Facts have attributes called slots like a name a status and so on Each kind of fact can have only a fixed set of slots The allowed slots for a given type are defined in Jess statements called deftemplates Actions like send late notice can be defined in user written functions in the Jess language deffunctions or in Java see chapter 15 For more information about rule syntax you can refer to
406. mend action check for loose RAM then continue You ll use the check fact as the trigger to enter the RAM flowchart shown later in figure 12 5 This trick lets the system ask further questions based on the results of a recommended action If you were writing a full diagnostic system with dozens of flowcharts you d use this trick many times Degrading gracefully If the computer is making a sound but the disk doesn t seek and there is no beep ing then the system has reached the limits of its knowledge represented by the oval labeled ERROR The unknown sound rule expresses this end point defrule MAIN unknown sound declare auto focus TRUE answer ident sound text yes answer answer gt halt recommend action ident seek ident does beep 12 5 3 To boot or not to boot If the disk does make a seeking sound then there is one more decision to make text no text no The rest of the rules 199 consult a human expert Does the computer begin to boot into the operating system before failing Two more rules cover the two possible answers defrule MAIN no boot start gt declare auto focus TRUE answer ident sound text yes answer ident seek text yes answer ident boot begins text no recommend action check keyboard RAM motherboard and power supply halt defrule MAIN boot start declare auto focus TRUE answer i
407. ment s ClassLoader is used to load any classes used by commands like new call and defclass public Rete Uses this as the application object Working with events These methods let you work with the Jess event system You can set the event mask to some combination of the constants in the JessEvent class and then register an instance of the JessListener class That instance will then receive all events that are generated public void addJessListener JessListener jel Registers a JessListener to receive events public void removeJessListener JessListener jel Unregisters a JessListener so it receives no more events public int getEventMask Gets the current value of the event mask public void setEventMask int mask Sets the event mask In general you should compose this value with the value returned by getEventMask using the operator B 8 2 B 8 3 B 8 4 jess Rete 421 Exchanging Java values with Jess language code These methods work with the same storage area accessed by the store and fetch commands in Jess By calling store in one language and fetch in the other you can easily move data back and forth between the two public void clearStorage Clears the storage area so that store returns null for any name public Value fetch java lang String name Retrieves an object previously stored with store public Value store java lang String name java lang Object val Stores an object under the given name
408. ment is given only facts whose head matches this symbol will be saved save facts does not work in applets the argu ment must be a valid path to a writeable local file If the filename contains special characters it must be passed as a double quoted string set lt object gt lt lexeme gt lt expression gt Sets a JavaBean s property to the given value The first argument is the Bean object the second argument is the name of the property The third value is the new value for the property the same conversions are applied as for new and call See get set current module lt lexeme gt Sets the current module Any constructs defined without explicitly naming a mod ule are defined in the current module Note that defining a defmodule also sets the current module set factory lt object gt Sets the thing factory for the active Rete object an instance of jess fac tory Factory Providing an alternate thing factory is a very advanced and cur rently undocumented way to extend Jess s functionality setgen lt number gt Sets the starting number used by gensym Note that if this number has already been used gensym uses the next larger number that has not been used set member lt object gt lt lexeme gt lt lexeme gt lt expression gt Sets a Java member variable to the given value The first argument is the object or the name of the class in the case of a static member variable The second argument is the
409. ments that contain bits of Java code A J2EE container knows how to extract that Java code translate the HTML into more Java code and compile the JSP into a servlet that emits the original HTML 19 5 2 Starting the Catalog servlet 333 plus the results of evaluating the Java code Thus a JSP is like the Hello servlet turned inside out Instead of being Java code with embedded HTML it s an HTML document with embedded Java code Unlike servlets JSPs are easy for web designers to edit Using JavaServer Pages as the interface to a web application while using servlets for the logic is an exam ple of using the right tool for the right job Here s a contrived example hello jsp that uses a Rete object to print the now familiar message Hello World from Jess lt html gt lt page import jess gt lt head gt lt title gt Hello World lt title gt lt head gt lt body gt lt H1 gt lt Rete engine new Rete engine addOutputRouter page out engine executeCommand printout page Hello World from Jess crlf gt lt H1 gt lt body gt lt html gt The variable out is the same as the return value of response getWriter from the Hello servlet and it s automatically available to all JSPs The important thing about hello jsp is that most of the file is plain vanilla HTML All the tags could be changed using just an HTML editor no Java programmer need be involved To install a JSP in Tomcat
410. mer and they ve never bought a rubber duck from us then recommend that they do so To test this rule you need several sets of facts one set with a customer who has bought a rubber duck and another set with a customer who has not You may also want to test a customer who has bought multiple rubber ducks You should also test using customers who have placed multiple orders in the past and customers who are new to the web site both duck buying customers and duckless ones 296 CHAPTER 17 Jess on the Web 17 5 2 The recommend requirements rule Perhaps the most important rule in the Recommendations Agent is the one that creates recommendations for products that are explicitly required by other prod ucts like the batteries for the television set mentioned earlier Because you re interested in using the Recommendations Agent to maximize revenue you won t recommend just any batteries for the television set You ll recommend the most expensive package in the catalog This rule is the longest one you ve written so let s break it down a little at a time The first two patterns match an order such that no other order with the same customer id has a higher order number that is it matches only the current order for any given customer defrule recommend requirements order customer id id order number currentOrder not order customer id id order number order2 amp gt Porder2 currentOrder The next pattern
411. method 308 Exchanging Java objects 308 Beyond executeCommand 310 Working with Fact objects in Java 311 Multislots 313 Ordered facts 313 Removing facts 314 Working with JavaBeans 314 Calling Jess functions from Java 315 Working with JessException 316 Nested exceptions 317 Rolling your own 318 Input and output 319 Using custom routers 320 Summary 321 xviii CONTENTS Deploying web based applications 323 19 1 19 2 19 3 19 4 19 5 19 6 19 7 19 8 19 9 19 10 19 11 19 12 19 13 19 14 19 15 19 16 The Java Servlet API 325 J2EE and the Tomcat engine 326 Deploying the Hello servlet 326 Your first Jess servlet 329 Deploying the Jess servlet 330 Cleaning up the URL 331 Application architecture take one 331 Starting the Catalog servlet 332 JavaServer Pages 332 Forwarding toa JSP 333 Application architecture take two 335 The login screen 335 The Catalog servlet 337 Initializing Jess 337 Getting the login name 338 Starting a user session 339 Querying the product list 340 Invoking the JSP 340 The catalog JSP 341 Testing 343 The Recommend servlet 343 Getting started 344 Creating the order 344 Getting the recommendations 345 Forwarding to JSPs 346 The recommend JSP 346 The Purchase servlet 348 Persistence 350 Deploying the application 351 What s next 352 Summary 353 Jess XML and the enterprise 357 20 1 Enterprise applications 358 What is the J2EE 359 What does that
412. metimes though methods don t return they throw exceptions instead In the next section you ll see how Jess handles exceptions Working with exceptions Java methods can signal an error by throwing an exception An exception is just a Java object and it s intended to be treated as a message from the failed code to the caller When a Java constructor or method throws an exception Jess receives or catches the message and makes it available to you Jess signals errors in your Jess code and failures in its own functions using exceptions too so this section is rele vant even when you aren t explicitly working with Java objects Working with exceptions 71 When Jess catches an exception in a Jess function or a Java method its default action is to print a detailed message including either one or two stack traces to the console If there is only one stack trace it shows where in Jess s own Java code the problem occurred If the exception occurred in a Java method you called from Jess a second stack trace pinpoints the error in the Java method Together these stack traces tell you exactly what happened in your Jess program Although these messages are useful to you the programmer they re generally not what you want to happen in a deployed system when a user would see them Therefore whenever you call a method that might throw an exception you should supply a handler to execute in response to the exception in place of Jess s
413. mming language it is likely that vendor specific rule languages will be with us for a long time Each rule language has its own strengths and weaknesses and the expressiveness elegance and power of a rule language can be a major factor in choosing an engine 2 6 Summary A rule based system is a computer program that uses rules to reach conclusions from a set of premises Its historical roots include production systems and expert systems but nowadays their broad range of applications includes everything from real time control of embedded systems to enterprise resource planning for multinational corporations Rule based systems are not procedural but declarative programs They require a different approach to programming in which a runtime system is used to make scheduling and control flow decisions Modern rule based systems often include hybrid procedural declarative languages broadening their applicability Awide range of commercial rule development and deployment environments is available but all have an essential architecture in common Efforts are under way to standardize rule engine APIs and rule programming languages These first two chapters provided an introduction to the fundamental con cepts of rule based programs This is really a practitioner s book however so we want to begin writing new rule based programs as soon as possible You will learn the Jess programming language in the next part of this book You ll start by l
414. mmunicate by calling Jess functions In the diagram time flows down Note how calling a Jess function like run results in a Java function like Rete run being called a short time later You ll adopt the following strategy to get the main thread and the event thread to cooperate 1 2 3 The main thread sets up the GUI to display the question The main thread sleeps until a reply is available When the user presses Enter or clicks the OK button the event thread asserts a fact containing a reply The main thread wakes up and processes the reply If the reply is invalid go back to step 1 If it is valid assert the answer fact and return from the module Jess contains a mechanism that makes this process simple to implement There is a function you can call to make the current thread sleep when no activated rules are on the agenda of the current module and then wake up and continue once a new activation arrives No Jess language function lets you access this mechanism directly however you have to call it as a Java method The class jess Rete Jess s rule engine is embodied in a class named jess Rete Many of the most important functions you ve seen so far in the Jess language run reset clear Getting answers 211 assert correspond to single Java method calls on instances of this class run reset clear assertFact When you start jess Main from the com mand line a single instance of jess Rete is created
415. more than 50 miles further from your old home than your old workplace With Form 2441 you can get credit for care for an elderly parent or other dependent not just care for children The IRS web site is at http www irs gov Alternatively you can get the forms directly at ftp ftp fed world gov pub irs pdf 160 9 3 3 9 4 CHAPTER 9 Collecting the knowledge Next steps You ve now amassed enough knowledge about the problem domain to write the application which you will begin to do in the next chapter You first need to orga nize the data by defining deftemplates and organize the rules by defining defmodules You also need to write some infrastructure functions for input and output for example In chapter 11 with the infrastructure in place you will write the rules and deploy the application Summary The application area for a rule based system is called its problem domain The pro cess of collecting information about a problem domain is called knowledge engineer ing Knowledge engineering can include gathering data from interviews books and other publications the Internet and other sources You ve begun work on an application that advises people about the forms they need to use to file their United States federal income taxes In this chapter you did the preliminary knowledge engineering and the end result is several lists of information chunks in prose form Designing the application In th
416. mories instead because it s easier to keep these names straight The notation LEFT p q RIGHT r indicates a test comparing the contents of slot qin the pth fact in a group from the left memory to the slot rin a fact from the right memory Join nodes produce one output for each ordered pairing of a left memory element and a right memory element that passes the tests in that node The oval nodes at the bottom of the network are the terminal nodes that repre sent individual rules They have a single input and no outputs When they receive an input they build an activation record from the input item and the rule they represent and place it on the agenda Note that any facts that reach the top of a join node could potentially contribute to an activation they have already passed all the tests that can be applied to single facts To run the network you present every new fact to each node at the top of the network The example pattern network eliminates all facts except the x y and z ones The join network then sends all x y pairs with x a y b to the termi nal node for example 1 and all x y z triples given the same restriction to the terminal node for example 2 The terminal nodes thus know what activation records to create What happens if after processing the initial facts we assert an additional fact z c 17 The fact is presented to the z pattern node and sent down to the join node below The left memory of that join node a
417. n Sometimes coming up with appropriate Readers and Writers is tricky In the jess Console graphical console application for example the output from printout t goes to one graphical text component in the GUI and readline t takes input from a second text component To make this work Jess includes two adapter classes named jess awt TextAreaWriter and jess awt TextReader TextAreaWriter is a subclass of Writer that takes a TextArea as a constructor argument The abstract write method is implemented to call append on the java awt TextArea accumulating text as it arrives The jess Console applica 18 7 Summary 321 write append TextAreaWriter read TextReader tA appendText Figure 18 1 Usage diagram for TextAreaWriter and TextReader tion installs a single instance of this class as the routers t WSTDOUT and WSTDERR As a result all of Jess s regular output goes to the TextArea see figure 18 1 The TextReader class is similarly used to collect user input from the GUI Text Reader is a subclass of Reader with read methods that return data from an inter nal buffer TextReader also has a method appendText that lets you add text to this internal buffer jess Console connects a java awt TextField in its GUI to an instance of TextReader using an event handler Whenever the user presses Enter in the TextField the event handler calls appendText on the TextReader making the text entered in the TextField availa
418. n A typical rule based system may contain hundreds or thousands of rules It s very likely that at any given moment more than one rule is activated The set of acti vated rules that are eligible to be fired is called the conflict set and the process of putting the rules in firing order is called conflict resolution The output of the con flict resolution process is the ordered list of activations called the agenda You can see this ordered list of activated but not yet fired rules with the agenda function Conflict resolution in Jess is controlled by pluggable con lict resolution strategies Jess comes with two strategies depth the default and breadth You can set the cur rent strategy with the set strategy command Using set strategy depth causes the most recently activated rules to fire first and set strategy breadth makes rules fire in activation order the most recently activated rules fire last In many situations the difference does not matter but for some problems the con flict resolution strategy is important Although the default strategy is intuitive and correct in most situations it runs into trouble if every rule that fires activates another rule The oldest activations then get pushed far down the agenda and never get a chance to fire The breadth strategy avoids this problem but the first in first out firing order can be confusing You can write your own strategies in Java by implementing the jess Strategy in
419. n rm f class If you re new to make note that the indented lines must start with real tab charac ters not spaces The install target in the Makefile compiles the application re creates the directory structure and installs all the necessary files This Makefile makes a good template for general servlet development you ll expand it and use it for the Recommendations Agent application later in this chapter Restart Tomcat The easiest and most reliable way to get Tomcat to load your new application or reload it after a change is to shut down Tomcat and start it up again You can use the startup and shutdown scripts in TOMCAT bin or you can do what I do and write a third script named restart that combines the two functions Look at the startup and shutdown scripts and create restart by analogy A properly laid out web application will be detected automatically by Tomcat The first time a browser request comes in for the Hello application Tomcat will load the web xml file to learn about the application and find out that a servlet named Hello is implemented in a class named Hello The first time the servlet is requested Tomcat loads it in and calls the doGet method you implemented to sat isfy the request The generated page is returned to the browser Your first Jess servlet 329 Seeing the results The default standalone Tomcat installation starts its own mini web server on port 8080 Fire up your favorite web browser
420. n engine These components are shown schematically in figure 2 1 3 W Van Melle A Domain Independent Production Rule System for Consultation Programs Interna tional Joint Conference on Artificial Intelligence 1979 923 925 20 CHAPTER 2 What are rule based systems Inference Engine fact f1 fact 2 A fact 3 Figure 2 1 The architecture of a typical a le based system The pattern Pattern Matcher u Working Memory matcher applies the rules in the rule base to the facts in working rule r3 memory to construct the 1 2 r1 TF agenda The execution engine 2 f3 r2 fires the rules from the agenda Agenda which changes the contents of working memory and restarts Rule Base the cycle Execution Engine El 2 rl 2 3 1 The inference engine If you wanted to write your own rule engine where would you start You might begin with the most important component The primary business of a rule engine is to apply rules to data That makes the inference engine the central part of a rule engine The inference engine controls the whole process of applying the rules to the working memory to obtain the outputs of the system Usually an inference engine works in discrete cycles that go something like this 1 All the rules are compared to working memory using the pattern matcher to decide which ones should be activated during this cycle This unordere
421. n its LHS and no actions on its RHS It will therefore always execute and it will accomplish nothing The following example uses two new arguments to the watch function activations and rules you used watch facts in chapter 6 Jess gt watch facts TRUE Jess gt watch activations TRUE Jess gt watch rules TRUE Jess gt reset gt f 0 MAIN initial fact gt Activation MAIN null rule f 0 Jess gt run FIRE 1 MAIN null rule f 0 1 The function call watch activations tells Jess to print a message whenever an activation record is placed on or removed from the agenda An activation record associates a set of facts with a rule It means the given set of facts matches the LHS of the given rule and so the rule should be executed In this case because nu11 rule doesn t specify a LHS Jess has automatically made it conditional on the pres ence of the initial fact You ll recall from chapter 6 that the reset function places a fact initial fact in working memory This is one important role for initial fact to serve as a trigger for rules with an empty LHS You can see the change using the ppdefrule function which pretty prints a rule by re creating its text from Jess s internal representation Jess gt ppdefrule null rule defrule MAIN null rule A rule that does nothing MAIN initial fact gt 98 CHAPTER 7 Writing rules in Jess The return value of ppdefrule is a string
422. n my machine Jess performance on the Miranker manners benchmark http www 2 cs cmu edu afs cs project ai repository ai areas expert bench bench 0 html with 90 guests is improved by 25 by adjusting the initial heap size and nursery size to 32MB and 16MB respec tively from their defaults of 64MB and 640KB In this case you make the default heap size smaller which makes the garbage collector run faster because there is less memory to search You can tune the HotSpot virtual machine in this way using the following command java XX NewSize 16m Xms32m Xmx32m jess Main lt scriptfile gt The object nursery is a subset of the Java heap set aside for recently allocated objects The total heap size in this example is 32MB not 48MB 3 3 Summary Jess is a powerful environment for processing rules and scripting the Java plat form You can use it in a wide range of applications built purely using the Jess rule language purely in Java or with some mixture of the two You can run Jess as an interactive command line application during develop ment using the jess Main class Jess programs can also be stored as plain text files and executed by jess Main See in particular http developer java sun com developer TechTips 2000 tt1222 html tip2 40 CHAPTER 3 Introducing Jess Jess works with any Java 2 virtual machine Jess is very fast but its perfor mance is sensitive to the detailed operation of the Java garbage collector
423. n turn for each value all the other expressions are evaluated in order The return function can be used to break the iteration This example uses foreach to print each item in the list foreach x createS a bcd printout t x crlf 400 APPENDIX A Jess functions format lt symbol gt lt string gt lt expression gt Creates a formatted string Formats the third and subsequent arguments accord ing to the format string the second argument which is identical to that used by printf in the C language refer to a C book for more information Returns the string and optionally prints the string to the I O router named by the first argu ment If you pass nil for the router name no printing is done gensym Returns a special unique sequenced value consisting of the letters gen plus an inte ger Use setgen to set the value of the integer to be used by the next gensym call get lt object gt lt symbol gt Retrieves the value of a JavaBean s property The first argument is the object and the second argument is the name of the property The return value is converted to a suitable Jess value exactly as for call You can refer to the JavaBeans specifica tion for the complete definition of a property but often a property X implies a method getX get current module Gets the current module see set current module get focus Returns the name of the current focus module see focus get focus stack Returns the mo
424. name is the name of the query and x y and so on are the variables named in the variables declaration for the query run query works by asserting 7 7 3 7 7 4 7 8 Summary 131 a fact to match this pattern using the arguments you supply to instantiate the vari ables This fact completes any pending matches of the defquery s LHS and run query collects these matches and returns them The count query results function To obtain just the number of matches for a query rather than a full Iterator over all the matches you can use the count query results function This function accepts the same arguments as run query but returns an integer specifying the number of matches Backward chaining and queries It can be convenient to use queries as triggers for backward chaining For exam ple look back at the backward chaining example in section 7 4 If you were writ ing a def function that needed to use factorials that deffunction might want to use a defquery to fetch the ones that are already available from working memory rather than recomputing them The backward chaining rules would then com pute missing values For this technique to be useful run must somehow be called while the query is being evaluated to allow the backward chaining to occur Facts generated by rules fired during this run may appear as part of the query results By default no rules fire while a query is being executed If you want to allow backward chaining
425. nction to catch that exception and return FALSE We looked at try in section 5 4 The rest of the rules Now that the infrastructure is squared away let s go back to looking at some rules If the computer whirrs beeps or makes any other sounds when turned on the top level flowchart directs you to continue on to the next flowchart shown in fig ure 12 4 Let s translate this into rules one piece at a time _sranr Go to Check fas loose RAM flowchart 1 Does NO Does disk computer seek beep YES 3 or more S Check ERROR keyboard motherboard Figure 12 4 Does No A simple flowchart for diagnosing bootin eae some hardware problems in egin 4 me computers that make sound but don t YES motherboard boot The ERROR state is supposed to power supply be impossible to reach enn Rules about sound The trigger to enter this flowchart in the first place occurs when the user answers no to the sound question therefore all the rules begin with the pattern answer ident sound text no Any decisions made before this point are irrelevant to these rules The first decision in the sound flowchart is whether the disk makes seeking sounds so similarly all the rules in this flowchart include a pattern to match one of the possible answers to the seek question From this point the rules diverge you can look at the path through t
426. nctions 393 create accepts a string the name of a new bag to be created The bag object itself is returned For example Jess gt bag create my bag lt External Address java util Hashtable gt delete accepts the name of an existing bag and deletes it from the list of bags find accepts the name of a bag and returns the corresponding bag object if one exists or nil a list returns a list of the names of all the existing bags set accepts as arguments a bag a String property name and any Jess value The named property of the given bag is set to the value and the value is returned get accepts as arguments a bag and a String property name The named property is retrieved and returned or nil if there is no such property For example Jess gt defglobal bag 0 TRUE Jess gt bind bag bag create my bag lt External Address java util Hashtable gt Jess gt bag set bag my prop 3 0 3 0 Jess gt bag get bag my prop 3 0 props accepts a bag as the single argument and returns a list of the names of all the properties of that bag Bag objects are local to each Rete object batch lt filename gt Attempts to parse and evaluate the given file as Jess code If successful returns the return value of the last expression in the file In an applet batch will try to find the file relative to the applet s document base In any program if the file is not found the name is then passed to ClassLoader g
427. nd set it to that value You also modify the init simulator function to allow you to specify the initial tem perature set point value This fuzzy controller behaves very well at all realistic outside temperatures and keeps the temperature extremely close to the set point As you can guess getting the right fuzzy values and the right rules can be complicated Should you have defined more fuzzy values for the rate of change Should the shape and position of the fuzzy set be adjusted to get the best performance moves quickly to the set point reduces overshoot of the set point and so forth A large body of work is devoted to determining these fuzzy control parameters and how they should 282 16 6 3 16 7 CHAPTER 16 Writing the rules change as the system changes dynamically Yet with a little common sense and the tools to describe your thoughts in an English like manner you have been able to construct a system that is robust to changes in outside temperature and set point and does an adequate job of keeping the temperature in a comfortable zone Exploring the fuzzy controller You might want to modify the second fuzzy controller to allow the set point as well as the outside temperature to be changed in the user interface Doing so should be relatively simple because the required hooks are already included in the code the method SetThermometerFVTerms in the Thermometer class You could make a number of other explorations Would def
428. ndations The rules you ve written so far can sometimes generate multiple recommenda tions for the same product You could complicate all the previous rules such that they won t generate the duplicate recommendations or you could simply allow them to be created and then clean them up at the end I ve chosen to take the lat ter route A single rule coalesce recommendations combines multiple recom mendations for the same product defrule coalesce recommendations If there are multiple recommendations for the same product coalesce them r1 lt recommend order number order type type because product part number part order number order part number part because products amp not members product products r2 lt recommend gt retract rl r2 assert recommend order number order type type because createS product products part number part The rule recommend requirements can generate another kind of invalid recom mendation It may recommend a product as a requirement even though the cus tomer is already buying another product in that same category If the customer is 17 7 Some useful queries 301 buying cheap batteries it would be pointless to recommend the expensive ones Again you could complicate the original rule to account for this situation or you could write a cleanup rule The rule remove satisfied recommendations retracts recommendations that a
429. nding Jess with functions written in Java you ll learn how to do this in part 5 of this book In chapters 6 and 7 you ll see how Jess s ability to work with Java objects comes into play in rule based programming as we study Jess s working memory and learn how to write Jess rules And as you ll see later in this book these capabilities make integrating Jess into a Java application extraordinarily easy Representing facts in Jess In this chapter you ll Explore Jess s working memory Learn about the different kinds of facts and when to use them Find out how to store Java objects in working memory 74 6 1 Jess s working memory 75 Now that we ve looked at the basic syntax of the Jess language it s time to begin our study of Jess s rule based programming features In this chapter we ll look at how to create and manage the data that rules can act on the working memory we first discussed in chapter 1 The working memory sometimes called the fact base contains all the pieces of information Jess is working with The working memory can hold both the premises and the conclusions of your rules Jess stores the contents of working memory using a set of customized indices that make looking up a particular piece of information very fast much as a rela tional database does Even though Jess uses a data centric index internally your view of working memory will look like a simple list Each item in working memory a
430. ng like this the actual output from compiling rule example 1 from section 8 3 1 MAIN example 1 14 1 1 2 t Each time 1 appears in this string it means a new one input node was created 2 indicates a new two input node t indicates a terminal node Now watch what happens when we compile example 2 from the same section MAIN example 2 1 1 1 1 2 2 t The string 1 is printed whenever a preexisting one input node was shared simi larly 2 means a two input node was shared You can see from these diagnostics that as expected one of the two join nodes in example 2 was shared along with most of the pattern nodes If you want to study what happens more precisely you can use the view command described in the next section The view function The view command see figure 8 4 is a graphical viewer for the Rete network By giving you feedback about the data structures Jess builds from your rule defini tions it may help you to design more efficient rule based systems Issuing the view command after entering the rules example 1 and example 2 from section 8 3 1 produces a very good facsimile of the diagram in figure 8 3 with some sub tle differences The various nodes are color coded according to their roles in the network Nodes in the pattern network are red normal join nodes are green not nodes are yellow and terminal nodes are blue The bottom node in the left col umn of the figure is a right to left adapter one of these is a
431. ng object Jess functions 407 retract string lt string gt Parses the string as a fact if such a fact exists in working memory calls retract on it return lt expression gt From a deffunction returns the given value and exits the deffunction immedi ately From the right hand side of a rule terminates the rule s execution immedi ately and pops the current focus module from the focus stack The argument is meaningless when return is called from a rule round lt number gt Rounds its argument toward the closest integer or toward negative infinity if exactly between two integers rules lt lexeme gt With no arguments prints a list of all rules and queries in the current module not the focus module to the t router With a module name for an argument prints the names of the rules and queries in that module With as an argument prints the names of all rules and queries run lt integer gt Starts the inference engine If no argument is supplied Jess keeps running until no more activations remain or halt is called If an argument is supplied it gives the maximum number of rules to fire before stopping run query lt lexeme gt lt expression gt Runs a defquery and returns a java util Iterator of the matches The first argument is the name of the query and the later ones are the arguments to the query Note that run query can lead to backward chaining which can cause rules to fire thus if run
432. ng rather than a symbol Doing so is necessary because you re matching a JavaBean property and JavaBeans can t have symbols as property values only strings Enough is enough If all the floors serviced by a heat pump are within the acceptable temperature range the heat pump should be shut off Exactly what should be the criteria for shutting off the pumps If the heat pump is in cooling mode there are three pos sible criteria You run the heat pump until too hot is not true for any floor You run the heat pump until the temperature on each floor is below the set point You run the heat pump until too cold is true for each floor The first case would lead to rapid cycling of the heat pump because when the temperatures went below the upper guard line the pump would be turned off and when they went above the upper guard line the pump would be turned on As I ve said rapid mode switching is bad for heat pumps so you don t want to do this In addition the average temperature would be the temperature at the guard line not near the actual set point The third case would lead to nice slow switching between the cooling and off states and an average temperature the midpoint between the lowest and highest operating temperatures equal to the set point The temperature would cycle between the upper and lower guard lines Unfortunately this means that at the moment the heat pump was switched off all the floors connected to i
433. ng the JavaBeans 231 Note that PropertyChangeSupport itself includes the addPropertyChangeLis tener and removePropertyChangeListener methods It would be nice if you could use PropertyChangeSupport as a base class for your JavaBean but you can t PropertyChangeSupport s only constructor accepts the source object the JavaBean as an argument and super this is invalid Java A default constructor that assumed the current object was the source would have been useful You can now add a run method to Thermometer that polls the temperature on a given floor sending out PropertyChangeEvents to notify Jess when the temper ature changes You ll also add a read only floor property to identify the individ ual Thermometer The result is shown in listing 14 3 Listing 14 3 The Thermometer class including automated notification package control public class Thermometer extends BeanSupport implements Runnable private Hardware m_hardware private int m_floor private double m_oldReading public Thermometer Hardware hw int floor m_hardware hw m_floor floor new Thread this start public int getFloor return m_floor public double getReading return m_hardware getTemperature m_floor public void run while true double reading getReading m_pcs firePropertyChange reading new Double m_oldReading new Double reading m_oldReading reading try Thread sleep 1000 catch I
434. ngine written in C was the original inspiration for Jess Jess and CLIPS were written by entirely different groups of people however and their implementations have always been very different Jess is dynamic and Java centric so it automatically gives you access to all of Java s powerful APIs for net working graphics database access and so on CLIPS has none of these facilities built in Still there is a strong similarity between the rule languages supported by these two systems Many of the core concepts of Jess were originally derived from CLIPS which was itself influenced by early rule engines like ops5 and ART NOTE FOR Jess s syntax is quite similar to CLIPS but Jess and CLIPS are different bite and unrelated systems Some Jess constructs defclass definstance defmodule have very different semantics in CLIPS whereas others de frule are virtually identical Jess has many features defquery the abil ity to directly call functions in the host language that CLIPS doesn t 3 1 1 3 1 2 The Jess rule engine 33 and Jess does not implement everything that CLIPS does COOL the CLIPS Object Oriented Language is one notable example If you have previous experience using CLIPS don t assume you can skip over this part of the book Obtaining Jess You can download a specially licensed version of Jess from the Manning Publica tions web site at http Awww manning com friedman hill The license lets you use
435. ngineering fields college textbooks are an excellent way to get an overview of a domain Introductory textbooks are often aimed at a general audience so you can read them without a specialized background The best textbooks have gone through several editions honing their language and presentation Monographs on specific topics can also be useful these are used as texts for advanced college and graduate level courses They are sometimes less well written and aimed at an audience with specific technical background Univer sity and technical libraries are a good source for textbooks and monographs Professional and scholarly journals are published several times each year and they are an excellent way to keep up with advances in a particular field They can be very expensive so you ll want to find them in a library as well Newsletters circulars and other publications aren t usually peer reviewed but they can provide useful information In particular many government publications 1 Or maybe not The U S General Accounting Office released a much publicized report in 2001 relating its findings that IRS telephone personnel give out incorrect tax information 47 of the time 9 3 9 3 1 Collecting knowledge about tax forms 157 are an invaluable way to learn about laws regulations and practices they com bine and distill information from various laws orders legal decisions and policies to produce practical guides Web sites and el
436. nk multifields too just use bare characters Matching global variables If you match to a defglobal with a pattern like score x the match only con siders the value of the defglobal when the fact is first asserted Subsequent changes to the defglobal s value will not invalidate the match if the rule was activated based on the value of the defglobal it stays activated even if the defglobal s value changes The match does not reflect the current value of the defglobal but only the value at the time the matching fact was asserted Connective constraints Quite often matching with a literal value or a variable isn t enough You might want a pattern to match if a client is located in either Boston or Hartford for example or you might want a pattern to match as long as the client is not from Bangor You can write these patterns and many others using the connective con straints amp and or and not Any single constraint preceded by a tilde matches the opposite of what the constraint would originally have matched For example the following pattern matches any client facts with a city slot that doesn t contain Bangor client city Bangor This pattern matches clients that have purchased exactly two items which must not be the same client items purchased x x 7 2 4 Constraining slot data 105 The other connective constraints let you form groups of single constraints Ampersands amp
437. nly two Java method declarations package jess public interface Userfunction String getName Value call ValueVector vv Context context throws JessException I ll discuss each of these methods quickly here and then explore each of them at greater length in the following sections 15 1 1 15 1 2 15 1 3 The Userfunction interface 239 The getName method The getName method should return the name Jess will use for the function Any legal Jess symbol see section 4 1 2 can be a function name including weird ones like lt and of course those are already taken The only important thing to remember about implementing getName is that it should always return the same name for a given object A Userfunction can t change its name over time Jess keeps Userfunction objects in a hash table using the name as the key so chang ing the name would corrupt this data structure A typical getName implementation looks like this public String getName return kill Some implementations are more complex see section 15 4 2 This Userfunc tion will be callable from Jess using the name kill that is ki11 The call method The most important part of a Userfunction is its call method Jess invokes the call method when your Userfunction is called from Jess language code The first argument to the function is a list of the arguments your function was called with and the second argument is an execution context
438. ns 289 Userfunction 238 258 260 281 315 318 arguments 240 call 239 getName 239 loading 239 state 246 users 155 V validate 207 validating input 194 value resolution 308 ValueVector 240 add 240 get 240 size 240 VARIABLE 422 variable constraints multiple 102 variables 47 declaring 48 external 130 global 48 internal 130 multifield 49 57 static 70 static member 69 top level 48 variableValue 243 VARIANT 241 vendor independence 386 vents 258 261 262 absolute positioning 269 relative positioning 276 verbosity 238 videotapes 293 view 144 413 Visio 189 Visitor pattern 419 W wait 211 waitForActivations 211 212 watch 77 97 295 315 413 all 170 201 calling from Java 315 compilations 143 watch activations 120 watch rules 120 web applications 295 308 324 326 persistence 350 testing 343 web architectures 288 web browser 289 web designers 335 341 and JSPs 333 web services 291 web users 303 web xml 327 330 338 352 servlet mapping 331 WEB INF 327 INDEX 443 weightedAverageDefuzzify 275 Western Illinois University 366 while 52 183 244 258 414 whitespace 42 wildcards 63 Windows 169 326 428 WinZip 33 working fluid 257 working memory 19 21 22 75 134 307 changes with time 136 implementation 314 similarity to RDBMS 81 World Wide Web 288 wrapper objects 232 Writer 320 WSTDERR 319 WSTDIN 319 WSTDOUT 319 WWW 156 X XCON 10 292 294 XML xxvii
439. ns aren t bad But if you need to work with code in a native library or interface to a legacy sys tem without an EJB interface or link in substantial amounts of existing Java code you may have a hard row to hoe to work around them Second programming with EJBs is complex The J2EE reference implementa tion comes with almost a dozen command line tools for managing parts of EJB 378 CHAPTER 21 Jess in the J2EE environment development commercial J2EE environments often provide the same functional ity via graphical interfaces There is a substantial market for expensive Enterprise Editions of Java development environments that include graphical tools for auto mating the EJB development process All the infrastructure costs something in performance as well Scalability the ability to perform as well for 1 000 clients as for a single one doesn t imply high performance Given all this a question suggests itself Should I be using EJBs That can be a hard question to answer but it boils down to this If you need many of the advanced services EJBs make available then yes you should use them If you re only considering EJBs because it s the thing to do then you re asking for trouble 21 1 4 Accessing external resources from EJBs If you decide to use EJBs you ll almost certainly find that you need to connect them to some external resources software that can t be directly integrated into the EJB environment O
440. ns the value of the last one progn lt expression gt The progn function is useful when you need to group multiple expressions together into one expression usually due to syntax restrictions of other functions as in the following example Jess gt bind n 2 2 Jess gt while progn bind n n n lt n 1000 do printout t n crlf 4 16 256 FALSE 54 4 3 5 4 3 6 CHAPTER 4 Getting started with the Jess language Of course this particular example could also be written more succinctly as follows Jess gt bind n 2 2 Jess gt while lt bind n n n 1000 do printout t n crlf However in many important cases progn is the only real alternative Many of these cases come up during pattern matching on the left hand sides of rules which you will see in chapter 7 The next few functions are used less frequently than the ones we ve covered so far in this section but when you need them you need them badly These func tions have no equivalents in Java they are powerful features available only in the Jess environment They are used to convert directly between data and code some thing that can t be done in a compiled language like Java apply In Java you write a different line of code to add two numbers than you would to multiply them print them or save them in a file In Jess you can write one line of code that depending on the values of some arguments can do any of these
441. nterruptedException ie return 232 14 3 1 CHAPTER 14 The reality connection The PropertyChangeSupport class needs the name of the property along with the old and new values to create the appropriate event Note that PropertyChangeSupport is smart enough to send an event only if the old and new values are different so you don t need to bother with that test yourself There are a few overloaded versions of the firePropertyChange method to handle dif ferent properties of different types For some types though like the double value here you need to use wrapper objects to pass the old and new values Rules about Thermometers Given this improved version of Thermometer it s possible to write rules that react to temperatures If this class and the Simulator class have been compiled and are available on the CLASSPATH the following code will print a warning message after a short interval Remember that all the code for the HVAC Controller is available from this book s web site Create a simulator bind hardware new control Simulator 3 Monitor the temperature on the first floor defclass Thermometer control Thermometer definstance Thermometer new control Thermometer hardware 1 Report when the temperature gets to be over 72 defrule report high temperature Thermometer reading r amp gt r 72 gt printout t It s getting warm on the first floor crlf halt run until halt exit
442. ntext servletContext getServletContext Rete engine Rete servletContext getAttribute engine String command get new order number int nextOrderNumber engine executeCommand command intValue null return nextOrderNumber You wrote the deffunction named get new order number in section 17 7 1 340 19 8 4 19 8 5 CHAPTER 19 Deploying web based applications Querying the product list After the customer logs in you want her to see a catalog of all the available products this is obviously oversimplified a real site would have a search capability because there would be more products than could be listed easily on one page The name of each product should be listed along with its catalog number price and a checkbox for ordering The Catalog servlet prepares the list of products and a JSP renders them as HTML Because you re using Jess as your product database the code to pre pare the list of products should simply run the all products query it looks like this ServletContext servletContext getServletContext Rete engine Rete servletContext getAttribute engine Iterator result engine runQuery all products new ValueVector request setAttribute queryResult result All this code does is to invoke the defquery and then store the java util Itera tor that comes back as an attribute queryResult in the request object where the JSP can find it Note that there are three different ways to s
443. nthe J2EE environment 375 A Jess functions 389 B Abridged Java API for Jess 415 C An automated testing framework 427 contents preface xxi acknowledgments xxiii about this book xxvi author online xxix about the title xxx about the cover illustration xxxi Rules to the rescue 3 1 1 Math class melee 4 Beyond logic puzzles 8 1 2 Some real world examples 9 Mail filtering 9 Product configuration 10 Implementing business rules 11 1 3 Summary 12 What are rule based systems 13 2 1 The cooking driving robot 14 Declarative programming a different approach 15 2 2 Rules and rule engines 17 Expert systems 18 CONTENTS 2 3 Architecture of a rule based system 19 The inference engine 20 The rule base 21 The working memory 21 The pattern matcher 22 The agenda 22 The execution engine 23 2 4 Developing rule based systems 23 Knowledge engineering 23 Structuring data 24 Testing 24 Interface building 25 Writing the rules 25 Iterative development 26 2 5 Rule engine standards 26 2 6 Summary 27 Introducing Jess 31 3 1 The Jess rule engine 32 Obtaining Jess 33 Installing Jess 33 Running Jess 34 Hello World 34 3 2 Jess applications 36 Command line GUI or embedded 37 Jess performance 38 3 3 Summary 39 Getting started with the Jess language 41 4 1 The basics 42 Whitespace 42 Symbols 42 The jess Value class 43 Numbers 43 Strings 44 Comments 44 4 2 Adding some
444. ntirely unrelated The and predicate function operates on Boolean expressions but the and conditional element oper ates on patterns You can always tell which you re dealing with by the context predicate functions can appear only as constraints on slot data The following are all of Jess s conditional elements m and Matches multiple facts or Matches alternative facts not Matches if no facts match m exists RMatches if at least one fact matches test Matches if a function call doesn t evaluate to FALSE m logical Matching facts offer logical support to new facts The and conditional element The LHS of every rule consists of a list of zero or more patterns Each of those pat terns must match for the whole LHS to match You might recognize this as the intersection operation from formal logic You can express the intersection of a group of patterns in Jess using the and conditional element The entire LHS of every rule is enclosed in an implicit and Any number of patterns can be enclosed in a list with and as the head The resulting pattern is matched if and only if all of the enclosed patterns are matched The following rule matches only if f laps up and engine on both match 7 3 2 Qualifying patterns with conditional elements 109 Jess gt defrule ready to fly and flaps up engine on gt Of course this rule would behave precisely the same way if the and CE was omitted so by itself and isn
445. ntity beans represent concrete objects or concepts like people or prod ucts Generally each entity bean corresponds to a single row in an underlying database so that changes to the bean are reflected automatically in persistent storage The EJB container can take care of this persistence automatically without your writing any database code If entity beans represent things session beans represent actions They contain business logic that coordinates the components of an application Stateful ses sion beans have short term persistence they are used for things like shopping carts If you restart the application server stateful session beans disappear whereas entity beans survive Stateless session beans merely encapsulate code and contain no client specific data Message driven beans are new with the EJB 2 0 specification They are stateless components that help integrate Java Message Service JMS services into the EJB container JMS is an asynchronous event based communication mechanism A quick tour of EJB concepts 377 21 1 2 EJB restrictions 21 1 3 To provide the services that it does an EJB container needs to maintain strict con trol over its internal state As a result there are certain prohibitions on what EJBs are allowed to do In general EJBs aren t supposed to do the following Use threads or synchronization Use the java io package to access files Display a GUI Receive incoming socket connections Set
446. ntroller xxvii HVAC controller xxiii 219 HVAC simulator 225 hyphen as word separator 97 hypothesis 188 191 216 I O routers 319 and binary I O 320 custom 320 330 IDE JessWin 259 if 401 Jess function 297 if then else 52 chaining 53 immutable 241 345 implode 401 import 63 204 401 63 index cards 152 162 inference engine 19 20 information overload 153 initial conditions 276 initial fact 77 79 97 98 112 114 142 143 init simulator 248 264 input validation 194 insert 401 instance variables 70 instanceof 401 Jess function 72 INTEGER 422 Integer 63 integer 196 402 Jess function 198 integerp 402 Intel x86 architecture 191 intelligent agents 37 interactive 76 interest income 174 175 178 interfaces building 25 Internet 288 Internet Explorer 289 interoperability 361 intersection 108 intersection 402 interviews 151 153 157 188 190 strategy 154 intranet 288 intValue 241 inventory 358 inventory management 288 IRS 150 156 159 is a function 60 INDEX 437 is of type 167 iterative development 26 162 Iterator 352 J J2EE xxvii 12 288 290 291 324 355 359 376 reference implementation 359 tools 377 J2EE Connector Architecture See JCA J2SE 384 Jakarta 324 359 JAR files 330 jargon 152 154 Java xxi arrays 67 214 calling methods from Jess 64 calling overloaded methods 68 calling static methods 66 compiler 68 constructing objects from Jess 62 downloading 33 perf
447. nts to the static showMessageDialog method are in order the par ent window the message the window title and a constant that indicates what kind of icon to display Another limitation of the application is that there s no way to exit You can tell the JFrame to call System exit method when it is closed using JFrame s setDefaultCloseOperation method frame setDefaultCloseOperation get member JFrame EXIT_ON_CLOSE Finally during testing I noticed that it s a bit disorienting to have question text remain in the window after the application has finished It would be nice to have the old question removed when the program is through You can t put code to do this into recommend action because sometimes this function is called to make an intermediate recommendation and more questions will follow You certainly don t want to go back and change all the rules to add GUI specific code What can you dor If you look at your rules you ll recall that they call halt as appropriate to ter minate execution of the system You can therefore use the defadvice function to modify the behavior of halt see section 4 5 to learn more about defadvice defadvice before halt qfield setText Close window to exit Now when a rule calls halt to halt execution the contents of the question field will automatically change Testing the interface Testing graphical applications automatically is difficult There are commercial to
448. ocket s new Socket www google com 80 use the socket catch Exception ex throw new JessException google Network error ex 18 6 Input and output 319 Any exceptions thrown by the Socket constructor result in a JessException being thrown the original exception is available via getNextException Input and output The function printout was the first Jess function you learned Until now the first argument to printout has always been t but I ve never explained why The time has come to break this silence and explain the topic of I O routers The following methods in the Rete class deal with these beasts addInputRouter String name Reader r boolean mode Defines or rede fines an input router getInputRouter String name Returns the named input router removeInputRouter String name Undefines an input router getInputMode String name Determines whether the named input router is console like a addoOutputRouter String name Writer w Defines or redefines an out put router getOutputRouter String name Returns the named output router removeOutputRouter String name Undefines an output router getOutStream Returns the standard output router a getErrStream Returns the standard error router You can use the readline and printout functions to collect input and display output from Jess printout t Enter y or n bind response re
449. of the sub sequent arguments see eq otherwise returns FALSE new lt lexeme gt lt expression gt Creates a new Java object and returns an EXTERNAL_ADDRESS value containing it The first argument is the class name java util Vector for example The new function looks in the table maintained by the import function to resolve class names that don t include a package name Any additional arguments are taken to be con structor arguments The constructor is chosen from among all constructors for the named class based on a first best fit algorithm as discussed in section 5 1 of the text Built in Jess types are converted as necessary to match available constructors not lt expression gt Returns TRUE if its only argument evaluates to FALSE otherwise returns FALSE nth lt integer gt lt list gt Returns the value of the specified field of a list value The first argument is the one based index of the field to return Jess functions 405 numberp lt expression gt Returns TRUE for numbers otherwise returns FALS EJ oddp lt integer gt Returns TRUE for odd numbers otherwise returns FALSE See evenp open lt lexeme gt lt lexeme gt r w a Opens a file the first argument names the file Subsequently the router identifier the second argument can be passed to printout read readline or any other functions that accept I O routers as arguments By default the file is opened for reading
450. of this function lies in the fact that you can call a function whose name for instance is in a Jess variable In this example a mathematical function is applied to a list of numbers Jess gets the name of the function from the console by calling read Jess gt bind function read You enter nye Jess apply function 1 2 3 4 5 15 asc lt lexeme gt Returns the Unicode value of the first character of the argument as an RU INTEGER assert lt fact gt Asserts all facts into working memory returns the fact ID of the last fact asserted or FALSE if no facts were successfully asserted for example if all facts given are duplicates of existing facts assert string lt string gt Converts a string into a fact and asserts it Attempts to parse the string as a fact and if successful returns the value returned by assert with the same fact Note that the string must contain the fact s enclosing parentheses Jess gt assert string grocery list milk bread soup lt Fact 0 gt bag lt command gt lt expression gt Lets you manipulate Java hash tables from Jess The net result is that you can cre ate any number of associative arrays or property lists Each such array or list has a name by which it can be looked up The lists can contain other lists as properties or any other Jess data type The bag command does different things based on its first argument It s really seven commands in one Jess fu
451. ok s web site http www manning com friedman hill Welcoming the user The startup module for this application is very simple It doesn t really have any work to do other than serving as a launching point You want to print a welcome banner when the user first sees the application so you can define a rule in startup to display one defmodule startup defrule print banner gt printout t Type your name and press Enter gt bind name read printout t crlt WKEKKK KKK KKK KKK KK KKK KKK KKKKKKA KEK crlf printout t Hello name crlf printout t Welcome to the tax forms advisor crlf printout t Please answer the questions and crlf printout t I will tell you what tax forms crlf printout t you may need to file crlf t printout WRK KK RR KR KEK KEK KKK KKK KKEKKE orif crlf Testing the startup module Whenever you write a rule or cooperating group of rules you should make sure they work To test the simple rule print banner you can just reset working mem ory set the focus and run text in bold indicates characters you type in response to a prompt Jess gt reset TRUE Jess gt focus startup MAIN Jess gt run Type your name and press Enter gt Fred KKK KKKK KKK KKK KKK KKK KKK KKK KKK KEKE KKEK Hello Fred Welcome to the tax forms advisor 174 CHAPTER 11 Writing the application Please answer the questions and I will tell you what tax forms you may need to file KKKKKK
452. olean systems This makes fuzzy rules a good choice whenever you need to make distinctions that aren t clearly drawn for example in systems that have to classify items into indistinct categories FuzzyJess written by Bob Orchard at the National Research Council of Can ada s Institute for Information Technology is a set of Jess extensions that gives you the ability to write fuzzy rules in Jess FuzzyJess is part of the Fuzzy Toolkit a powerful set of fuzzy logic tools written in Java In the remainder of this chapter we ll show how to use Fuzzy and FuzzyJess to write a fuzzy logic version of the HVAC Controller system Fuzzy logic briefly In the real world there exists much fuzzy knowledge knowledge that is vague imprecise uncertain ambiguous inexact or probabilistic in nature Human thinking and reasoning frequently involve fuzzy information possibly originating from inherently inexact human concepts and matching of similar rather then identical experiences In systems based on a classical set theory and two valued logic it is very difficult to answer some questions because they do not have com pletely true or false answers Humans however can often give satisfactory answers to these questions Fuzziness Fuzziness occurs when the boundary of a piece of information is not clear cut For example words such as young tall good and high are fuzzy No single quantitative 1 You can get FuzzyJess and the Fuzzy
453. ollowing it otherwise returns FALSE lt gt lt number gt lt number gt Returns TRUE if the value of the first argument is not equal in value to all subse quent arguments otherwise returns FALSE lt number gt lt number gt Returns TRUE if the value of the first argument is equal in value to all subsequent argu ments otherwise returns FALSE The integer 2 and the float 2 0 are but not eq gt lt number gt lt number gt Returns TRUE if the value of each argument is greater than that of the argument following it otherwise returns FALSE gt lt number gt lt number gt Returns TRUE if the value of each argument is greater than or equal to that of the argument following it otherwise returns FALSE abs lt number gt Returns the absolute value of its only argument agenda lt module name gt Displays a list of rule activations to the WSTDOUT router If no argument is speci fied the activations in the current module not the focus module are displayed Ifa module name is specified only the activations in that module are displayed If is specified then all activations are displayed 392 APPENDIX A Jess functions and lt expression gt Returns TRUE if none of the arguments evaluate to FALSE otherwise returns FALSE apply lt expression gt Returns the result of calling the first argument as a Jess function on all the remaining arguments The strength
454. ols to help you do it but most of them are fragile hard to use and expensive The classic problem with automated GUI testing tools is that they make your life harder rather than easier when you need to make changes to your interface because changes to the GUI may force changes to the test It s easy to write tests for a GUI that are so fragile that they must be rewritten any time the GUI layout changes even a little Tests that expect particular GUI components to lie at specific 216 13 8 CHAPTER 13 Adding a graphical interface screen coordinates are the most delicate in this respect Ifyou do use a GUI testing tool try not to write tests based on absolute screen coordinates Tests that refer to components by name are much more robust You can use the Java class java awt Robot to write your own automated tests This class lets you move the mouse pointer press mouse buttons and type text as if from the keyboard all under programmatic control I ve personally found the combination of java awt Robot and Jess s scripting capabilities to be particu larly powerful Besides automated testing you can of course do manual testing If as was the case here you wrote and tested the underlying logic of the application first test ing the GUI by hand isn t quite so bad Rather than trying to test every path through the knowledge base your GUI testing effort should be concentrated on trying every situation the GUI is meant to handle For
455. on This code loops using foreach over the four names given in the problem and creates using assert a pants color fact for each of the possible name color pairs and a position fact for each name position pair for a total of 32 facts The function create returns a list of its arguments Now that you ve written a rule to create all the possible combinations you ll write a second rule to search through them to find the subset of facts that repre sent the solution This is the fun part You ll translate each sentence in the prob lem statement directly into code First note that you use a symbol starting with a question mark like c to write a variable in Jess You ll use the variable c to CHAPTER 1 Rules to the rescue 6e 77 218 oe represent some color p to represent some position n to mean some name and c1 c4 and pl p4 to represent Fred Joe Bob and Tom s pants color and position respectively Here s the first useful sentence The golfer to Fred s immediate right is wearing blue pants defrule find solution There is a golfer named Fred whose position is pl and pants color is cl position of Fred is pl1 pants color of Fred is c1 The golfer to Fred s immediate right is wearing blue pants position of n amp Fred is p amp eq p pl 1 pants color of n amp Fred is blue amp c1 In this code snippet the variable n represen
456. on the login screen if none return to the login screen 2 Call get new order number to get a unique order number 3 Create a new user session with this login name and number 4 Use the all products query to get a list of product facts 5 Forward the results to the JSP for display Let s look at the Catalog servlet to see how these steps are implemented Ill show the source code for Catalog a little at a time if you find this confusing the full source code is available on this book s web site 19 8 1 Initializing Jess The Catalog servlet gets right down to business with its doGet method public class Catalog extends BaseServlet public void doGet HttpServletRequest request HttpServletResponse response throws IOException ServletException checkInitialized The web application uses a Rete object as its product database Where does the Rete object come from For efficiency all customers will share a single Rete object you can keep their data separate by using distinct order numbers You want the Rete object to be initialized when the web application is first started and you want it to persist as long as the web server is running You can write a single method that checks whether the Rete object has been initialized and creates and initializes a new one if not The Rete object should be stored in a location that is shared among all the servlets in the application You can then call this method call it checkInitialized fro
457. onditional element When you turn on your kitchen faucet you expect water to come out if it doesn t you ve got a plumbing problem When you turn off the faucet the flow Qualifying patterns with conditional elements 115 of water stops as a result This kind of relationship is called a logical dependency the water flowing is logically dependent on the faucet being open To express this idea in Jess you could write the following two rules Jess gt defrule turn water on faucet open gt assert water flowing TRUE Jess gt defrule turn water off not faucet open water lt water flowing gt retract water TRUE Given these two rules asserting faucet open will automatically cause water flowing to be asserted as well and retracting faucet open will retract water flowing if you call run so the rules can fire of course The fact water flow ing can therefore be said to be logically dependent on faucet open Writing two rules to express the one idea of logical dependency gets the job done but there is an easier way The logical conditional element lets you specify these logical dependencies among facts more concisely All the facts asserted on the RHS of a rule are logically dependent on any facts that matched a pattern inside a logical CE on that rule s LHS If any of the matches later become invalid for instance because one of the facts is deleted the dependent facts are retract
458. ontext context Value num vv get 1 resolveValue context double frac num floatValue context num intValue context Note that the variableValue method returns the name of the variable actually passed and Context has a public method setVariable that you can use to set a variable s value by name This suggests that Jess s built in bind function could sim ply be written as follows package jess class Bind implements Userfunction public String getName return bind public Value call ValueVector vv Context c throws JessException Value rv vv get 2 resolveValue c c setVariable vv get 1 variableValue c rv return rv And in fact it is Note that because ValueVector get will throw an exception if there aren t enough arguments and variableValue will throw an exception if vv get 1 isn t a variable this function like many Userfunctions doesn t have to do any explicit error checking Resolving function call arguments What happens if ki11 is called like this Jess gt kill 1 1 What is the type of the first argument in this case It is an RU FUNCALL value a representation of the function call In many languages Java is just one example all the arguments to a function are fully resolved before the function is invoked Jess however works differently none of the arguments is resolved before you see them Just as for variables function call arguments are reso
459. oop to render the rows of the table Each row contains some text cells plus an HTML INPUT widget of type checkbox Each checkbox is labeled with the part number of the corre sponding product 342 CHAPTER 19 Deploying web based applications lt while queryResult hasNext Token token Token queryResult next Fact fact token fact 1 String partNum fact getSlotValue part number stringValue null gt lt TR gt lt TD gt lt fact getSlotValue name stringValue null gt lt TD gt lt TD gt lt partNum gt lt TD gt lt TD gt lt fact getSlotValue price floatValue null gt lt TD gt lt TD gt lt INPUT type checkbox name items value lt partNum gt gt lt TD gt lt TR gt lt gt All that is left is to provide a button to submit the form and a bunch of closing tags This part is again plain HTML lt TABLE gt lt INPUT type submit value Check Order gt lt FORM gt lt BODY gt lt HTML gt Figure 19 3 shows what the rendered page looks like in a browser ttp Localhost 8080 Order cataleg AG Tekmart com Catalog Select the items you wish to purchase and press Check Order to continue Name Catalog Price Purchase TekMart VC 103 VCR TMVC103 1250 09 TekMar DVD 2z3 DVD Player TMDVDzzs zou 0 TekMart TV 13 19 TV mrvis 2z30 9 TekMart TV 24 24 TV MmTv24 2990 0 E
460. or or method Jess has the java lang Class object that represents the for mal parameter s type and a jess Value object that contains the value you passed and wants to turn the Value s contents into something assignable to the type named by the Class Hence the symbol TRUE could be passed to a function expecting a boolean argument or to one expecting a String argument and the call would succeed in both cases 64 5 2 CHAPTER 5 Scripting Java with Jess Table 5 1 Standard conversions from Jess types to Java types Jess type Possible Java types RU EXTERNAL_ADDRESS The wrapped object The symbol nil A null reference The symbol TRUE or FALSE String java lang Boolean or boolean RU ATOM a symbol RU STRING String char or java lang Character RU FLOAT float double and their wrappers RU INTEGER long short int byte char and their wrappers RU LONG long short int byte char and their wrappers RU LIST A Java array You ve created a HashMap but you haven t done anything with it Let s turn it into a lookup table by filling it with grocery price data Calling Java methods If you have a reference to a Java object in a Jess variable you can invoke any of that object s methods using the call function Let s work with the HashMap you created in the previous section HashMap put associates a key with a value and HashMap get lets you look up a value by key In th
461. or that allows you to customize the display for rule editing would be easier to use Some rule engines come with rule editors and others come with toolkits that let you create your own custom rule editor Current versions of Jess don t come with a rule editor but you can write your own custom editor in Java Using an Rule editors 373 XML rule format simplifies this approach quite a bit because you don t need to worry about parsing input files or writing output files standard XML libraries handle this for you easily For example to load an XML file as a tree of Java objects a DOM tree you can use the following import java io import org w3c dom import org xml sax InputSource import javax xml parsers PY sks public Document readInRules String filename throws Exception DocumentBuilderFactory dfactory DocumentBuilderFactory newInstance DocumentBuilder docBuilder dfactory newDocumentBuilder FileReader reader new FileReader filename try return docBuilder parse new InputSource reader finally reader close You can then use the interfaces in the org w3c dom package to examine and mod ify the DOM tree To write the modified DOM tree back out to disk you can use the following import javax xml transform import javax xml transform dom DOMSource import javax xml transform stream StreamResult ya eee public void writeOutRules Document rules String filename throws Excep
462. ormance 38 public member variables 70 required knowledge 32 version 33 Java Naming and Directory Inter face See JNDI Java Native Interface See JNI Java objects sending to Jess 309 Java Plug in 289 Java Servlet API 325 Java Web Start 289 java applet Applet 289 java awt Color 309 java awt Point 70 java awt Robot 216 java beans 230 java beans Introspector 88 java beans PropertyChange Event 92 438 INDEX java beans PropertyChange Listener 92 java io Reader 319 java io Serializable 247 java lang 63 java lang Object wait 211 java lang System exit 215 java lang Thread 209 java rmi Remote 380 java rmi RemoteException 380 java util ArrayList 425 java util Iterator 129 java util Vector 425 JavaBeans 66 87 88 229 258 260 314 example 89 properties 88 229 property names 67 string properties 260 JavaScript 292 JavaServer Pages See JSPs javax rules 26 353 383 reference implementation 386 javax rules admin RuleAdminis trator 385 javax rules RuleExecutionSet 385 javax rules RuleExecutionSet Provider 385 javax rules RuleServiceProvider 384 javax rules RuleServiceProvider Manager 384 javax servlet Servlet 290 javax swing 204 JBoss 359 JButton 204 207 JCA 378 384 JComboBox 67 68 213 JDBC 385 JDK 289 Jess xxi 18 applications 36 as a dynamic scripting envi ronment 37 as a prototyping tool 62 as a rapid application devel opment environment 37 built in functions 46 command line interface 35
463. orthy people 2 person age x amp lt x 30 gt printout t x is under 30 crlf Which form you use here is mostly a matter of taste I tend to use the test CE only for long or complex functions that would be hard to read if they were written as predicate constraints Of course the test CE can also be used to write tests that are unrelated to any facts import java util Date defrule fire next century test new Date after new Date Dec 31 2099 gt printout t Welcome to the 22nd century crlf For rules like this in which a test CE is the first pattern on the LHS or the first pattern in an and CE or the first pattern in any branch of an or CE Jess inserts the pattern initial fact to serve as the preceding pattern for the test The fact initial fact is therefore also important for the proper functioning of the test conditional element the caution about reset in the preceding section applies equally to test The rule fire next century won t fire until reset is called after the twenty second century begins The test and not conditional elements may be combined so that not test eq x 3 is equivalent to test neq x 3 The conditional elements we ve looked at so far affect how a rule matches work ing memory There is one conditional element we haven t covered yet and it s unusual in that instead of affecting how a rule matches it affects what happens when a rule fires The logical c
464. ou can build applications with it I hope that what I ve described in this chapter can serve as a sourcebook when you design your own rule based J2EE applications and that I ve brought out some of the important issues After an overview of Enterprise JavaBeans we looked at building a custom rule server using Jess and Java RMI Servers like this one are in widespread use You can customize this simple rule server for your own needs by adding more methods or by adding an object pool or other lifecycle services The javax rules API which as of this writing is still being developed prom ises to provide portable generic access to rule engines from multiple vendors The API includes sophisticated features for managing rule bases in the J2EE envi ronment but is still fairly simple to use in runtime components Jess in Action has taught you to program in the Jess language and build a wide variety of rule based software From the simple command line Tax Forms Advisor to the sophisticated web based Recommendations Agent from the interactive PC Repair Assistant to the autonomous HVAC Controller you ve seen how rule tech nology can be applied to many different kinds of applications Each of the systems youve developed in this book can serve as a launching point for a whole family of software and still you ve barely scratched the surface of the practically unlimited potential of Jess Jess functions 390 APPENDIX A Jess functions Thi
465. ou can see that one join node is performing exactly the same function integrat ing x y pairs in both rules you can share that also as shown in figure 8 3 Sharing join nodes is an especially fruitful optimization Because joining involves compar ing facts to one another the tests in join nodes tend to be executed many more times than those in pattern nodes so much so that the time spent in the join net work generally dominates the running time of the system By sharing this one join node then you ve effectively doubled the performance of your program The pattern and join networks in figure 8 3 are collectively only half the size they were in the first version This kind of sharing comes up frequently in real sys tems and is a significant performance booster oo LEFT 0 a RIGHT b example 1 N Figure 8 3 A Rete network that shares both example 2 pattern and join nodes Performance of the Rete algorithm 141 8 5 Performance of the Rete algorithm 8 5 1 The Rete algorithm s performance is even harder to analyze precisely than the simple rules finding facts algorithm In general the performance on the first cycle is basically the same for the two algorithms Rete has to do pattern matching for every fact in working memory because there are no previous results to con sult In the worst case where every fact changes on every cycle and there is no net work sharing between rules then the perform
466. ou might be tempted to use lists to build trees or other nested data structures Plain lists can t be nested however Jess will flatten out any nested plain list you attempt to create Jess gt bind more groceries create grocery list salt soap eggs bread milk salt soap Usually when you have a grocery list you walk through a grocery store find each item on the list and put it into your shopping cart This involves iterating over the list performing an action for each item of the list To do this you need to learn about control structures in Jess Control flow Up until this point the code you ve written executes sequentially one statement at a time It s hard to write interesting programs this way Luckily the Jess lan guage has a number of functions that affect the flow of control and let you group statements make decisions and execute loops 4 3 1 Control flow 51 In describing these control flow functions I ll use the term expression An expression is any Jess value a symbol number string variable or function call To evaluate an expression means to use the value of the constant expression take the value that the variable represents or evaluate the function and use its result We ll discuss the following control flow functions in this section m apply Calls a function on a given set of arguments m build Parses and executes a function call from a string eval Parses and executes a function ca
467. ou need to do is intercept any call to nths and add one to the first argument before nths sees it That way when you call nth you can pass zero based indices but nths will see one based indices Imagine that the variable argv holds a copy of a function call to nths using a zero based index that is it s a list like this nths 0 grocery list You need to write some Jess code to change this list to use a one based index nths 1 grocery list The code to make this change looks like this Strip off the function name bind tail rest argv Convert zero based to one based index bind index 1 nth 1 tail Put the argument list back together bind Pargv create S nth index rest tail You saw the rest function in section 4 2 4 It strips the first element from a list and returns the remainder You needed to use nth to write this code and you used a one based index because that s what nth wants by default Now you can use this block of code to modify the behavior of nths In the body of a defadvice call the variable argv is special It points to the actual function Fine tuning a function s behavior 59 call being executed By modifying the contents of argv you can modify the argu ments the built in nths function sees First let s watch the default behavior of nths in action Jess gt bind grocery list create peas beans butter milk peas beans butter milk Jess gt nth
468. our approach to knowledge engineering After talking to the customers you should 152 CHAPTER 9 Collecting the knowledge have a rough idea of what the system should do and how long development is expected to take Next it s best to seek out general resources you can use to learn about the fun damentals of the domain and do a bit of self study Being at least vaguely familiar with the jargon and fundamental concepts in the domain will let you avoid wast ing the time of people you interview later You should learn enough about the fundamentals to have a rough idea of what kinds of knowledge the system needs to have Once you ve developed an understanding of the basics you re ready to begin the iterative process Based on your initial research write down a list of questions about the domain which if answered would provide knowledge in the areas you previously identified Seek out a cooperative subject matter expert briefly explain the project to him and ask him the questions often the customer will provide the expert otherwise they should pay the expert a consulting fee to work with you Usually the answers will lead to more questions After the initial interview you can try to organize the information you ve gath ered into some kind of structure perhaps a written outline or a flowchart As you do this you can begin to look for what might turn out to be individual rules For the Tax Forms Advisor an individual rule you mi
469. out bacterial cultures In chapter 13 you ll package the application with a graphical user interface suitable for use by office personnel You ll use Jess s powerful Java scripting capa bilities to build a Swing based graphical interface without writing any Java code The interface hooks into an enhanced version of the ask module so in adding the interface you won t need to modify the knowledge base you develop here Using flowcharts in knowledge engineering Diagnostic knowledge can be represented fairly well using flowcharts A lot of infor mation can be compactly represented by a few symbols as shown in table 12 1 A simple flowchart for dealing with automotive problems is shown in figure 12 1 Table 12 1 Symbols used in flowcharts Symbol Meaning Example Square box A simple action Repairing or replacing a component Diamond box A choice or decision Is the screen dark 1 R Davis B G Buchanan and E H Shortliffe Production Systems as a Representation for a Knowl edge Based Consultation Program Artificial Intelligence 8 1977 15 45 12 1 1 Using flowcharts in knowledge engineering 189 Table 12 1 Symbols used in flowcharts continued Symbol Meaning Example Oval box Beginning or end of process START Arrows Flow between boxes What to do for each branch of a decision If you re going to be drawing a lot of flowcharts a software program like Visio can help When
470. out t No such service at host crlf type print else if instanceof ERROR MalformedURLException then custom printout t Address has a syntax error crlf message else if instanceof ERROR AccessException then printout t You don t have permission crlf else if instanceof ERROR RemoteException then printout t Network error crlf else printout t Unknown Error crlf Default error No such service at host message Summary 73 Jess s throw function lets you throw Java exceptions from Jess code It works just like the throw keyword in Java The single argument must be an instance of a Java class that extends java lang Throwable Here s an example Jess gt throw new Exception Testing Jess reported an error in routine throw while executing throw new Exception Testing Message Exception thrown from Jess language code Program text throw new Exception Testing at line 1 Nested exception is java lang Exception Testing at jess TryCatchThrow call Funcall java 827 at jess FunctionHolder call FunctionHolder java 37 5 5 Summary Jess can create Java objects call their methods and access their data Jess can also work with Java primitives by converting between Java and Java types With a few exceptions particularly working with large or multidimensional arrays most Java code can be directly translated to Jess Even these exceptions can easily be overcome by exte
471. own applications What s next Now that you have built the HVAC Control system you may have decided that it s overkill After all mechanical thermostats do a perfectly good job of controlling 16 8 Summary 283 the temperature of most houses and many larger buildings too The beauty of the rule based controller is that it is easily extensible to cover many complicated situations Do you want the set point to be lower on winter weekends and higher on summer weekends Should the temperature be lower or higher at night to save energy or should the system simply be turned off in the temperate spring and autumn It would be easy to add a fancy rule based timer to the program a Date JavaBean that broadcasts regular time and date updates would be all the infrastructure you would need Rules to change the set point based on the date and time would then be trivial Would you like the occupants of the building to be able to set the temperature of their own floor Again the modifications would be simple You might use set point facts instead of a single defglobal to track the separate set points and modify the rules to match these facts instead of referencing set point Although a rule based solution to this control problem may seem more com plex than traditional alternatives the complexity is accompanied by greatly increased features and flexibility Summary In part 5 of this book you ve written a powerful system capabl
472. owser behavior varies widely between brands or even between releases of a single browser The Java Plug in was designed to mitigate this problem but getting poten tial users to install this hefty download is still an issue especially on the Internet Thin client applications In a thin client architecture most of the application specific code runs on the server A web based thin client is usually just HTML in a browser There are many ways to write thin client web applications in Java Although I will describe them here one by one they are often used in combination Serviets Servlets the server side analogue of applets are small modules that a web server or other container invokes in response to browser requests Just as an applet is a sub class of java applet Applet a servlet is an implementation of the javax serv let Servlet interface or most commonly a subclass of the class javax serv let http HttpServlet Whereas applets run inside a browser on the user s desktop servlets run in a container on a server machine The Java 2 Enterprise Edi tion J2EE is a standard for deploying applications on a server it includes among other things a servlet container Java architectures for the Web 291 The HttpServlet class includes methods you can override to process client requests that come in over the Web It also provides methods the container invokes to initialize and destroy the servlet You ll develop several servlets in chapter 1
473. pattern in this rule matches a shopping cart fact with any number of values in the contents multislot defrule any shopping cart shopping cart contents items gt printout t The cart contains items crlf The pattern in this rule matches any shopping cart fact with a contents slot that contains milk preceded by any number zero or more of items and followed by any number of additional items defrule cart containing milk shopping cart contents before milk S after gt printout t The cart contains milk crlf 104 7 2 3 CHAPTER 7 Writing rules in Jess As shown here multifields are accessible on the RHS of the rules that use them in patterns just as normal variables are A multifield always contains the matched values as a list even if it matches zero or one value You can and generally should as a matter of style leave the sign off a multifield when you refer to it on the RHS of a rule because there it is acting as a normal variable Blank variables You can match a field without binding it to a variable by omitting the variable name and using a question mark as a placeholder This is generally only useful as a way to specify that a multislot contains a certain arrangement of values with out caring what those values are For example a pattern like poker hand ten ace matches any poker hand starting with a ten ending with an ace and con taining a total of five cards You can have bla
474. pilations command 143 The view function 144 The matches function 146 8 8 Summary 146 Collecting the knowledge 149 9 1 The Tax Forms Advisor 150 9 2 Introduction to knowledge engineering 150 Where do you start 151 Interviews 153 Desk research 156 9 3 Collecting knowledge about tax forms 157 An interview 157 Reviewing the forms 159 Next steps 160 9 4 Summary 160 Designing the application 161 10 1 Organizing the data 162 10 2 Filling in details 163 Default slot values 163 10 3 More templates 164 10 4 Templates you don t need 165 10 5 Organizing the rules 166 10 6 Building the infrastructure 166 Simple text based I O 167 Fetching the question text 168 10 7 Summary 170 Xiv CONTENTS Writing the application 172 11 1 Welcoming the user 173 Testing the startup module 173 11 2 Asking the user questions 174 Income and dependents 175 Dealing with special circumstances 176 Testing the interview module 177 11 3 Recommending forms 177 11 4 Explaining the results 181 Testing the report module 182 11 5 Finishing touches 182 11 6 Testing the full application 183 11 7 Summary 184 Writing the PC Repair Assistant 187 12 1 Using flowcharts in knowledge engineering 188 From flowcharts to rules 189 12 2 The problem domain 190 Writing the first rules 191 12 3 Asking questions with backward chaining 193 12 4 Checking the answers 194 Modifying the ask module 195 12 5 The rest ofthe rules 197 Ru
475. plicitly by defining a new rule with the same name as an existing one every subrule associated with that rule is also removed A note regarding subrules and efficiency You ll learn in chapter 8 that similar patterns are shared between rules in the Rete network avoiding duplicated com putation Therefore splitting a rule into subrules does not mean the amount of pattern matching work is increased much of the splitting may indeed be undone when the rules are compiled into the network On the other hand keep the implementation in mind when you define your rules If an or conditional element is the first pattern on a rule all the subsequent pattern matching on that rule s LHS won t be shared between the subrules because sharing occurs only as far as two rules are similar reading from the top down Placing or conditional elements near the end of a rule leads to more shar ing between the subrules and thus more efficient pattern matching Finally I should mention that although subrules will probably always be part of the implementation of the or conditional element in Jess it is likely that they will no longer be user visible at some time in the future The not conditional element You may have heard the saying two wrongs don t make a right when you were growing up How can the opposite of the opposite of something not be the same as the original thing Well as it turns out it s quite often not Such is the case in real worl
476. ppears on this list in the order in which it was added We ll begin our study of working memory by looking at the individual items it contains the working memory elements or facts Jess s working memory Jess maintains a collection of information nuggets called facts This collection of facts is known as the working memory All the pieces of information your rules work with are represented as facts in the working memory For example if your pro gram is a smart environmental control system for an office building the facts in the working memory may be temperature and humidity readings from around the building and sensor readings and switch settings from the building s air con ditioning systems see figure 6 1 The contents of Jess s working memory are held in your computer s RAM Facts come in different types However like most other constructs in Jess facts are stored as lists Each of the following is a valid fact Figure 6 1 Facts in the working memory are Jess s representation of temperature floorl 20 5 Ko NI information about the temperature floor2 21 3 outside world humidity floorl1 55 humidity floor2 61 Working Memory 76 6 1 1 CHAPTER 6 Representing facts in Jess initial fact little red Corvette groceries milk eggs bread As it happens these are all examples of ordered facts which will be
477. query is called on a rule s right hand side other rules right hand sides may run to completion before the instigating rule completes Putting run query on a rule s right hand side can also cause the count of executed rules returned by run to be low Calling run query on the left hand side of a rule can cause strange effects be sure you know what you re doing Note that the Iterator returned by this function should be used immediately It will become invalid if any of the following functions are called before you ve used it reset count query results or run query It may become invalid if any of the following are called assert retract modify or duplicate and any of the affected facts are involved in the active query s result Each match is a jess Token object as described in section 7 7 of the text 408 APPENDIX A Jess functions run until halt Runs the engine until halt is called Returns the number of rules fired When there are no active rules the calling thread is blocked waiting on the engine s activation semaphore until new activations become available This function is typi cally called either as the last expression in a script or from its own thread save facts lt filename gt lt lexeme gt Saves facts to a file Attempts to open the named file for writing and then writes a list of all facts in working memory to the file This file is suitable for reading with load facts If the optional second argu
478. r example the pattern hasa x snout would look like this in RuleML lt fact gt lt atom gt lt _opr gt lt rel gt hasa lt rel gt lt _opr gt lt var gt x lt var gt lt ind gt snout lt ind gt lt atom gt lt fact gt This means that rule editors and in general all rule processing software can work with this kind of pattern without knowing anything about the meaning of the data The metadata about the pattern the pattern names slot names and so on is explicit indeed it s the important part This approach is powerful in its general ity The downside to this kind of rule language however is that it is verbose and that some of the software needed to work with it is fairly complex An alternate approach is to represent the specifics of the pattern not as data but in the elements Many people use this approach in specialized domains to good effect for example Laurence Leff of Western Illinois University has pub lished extensively on using such an approach with XML representations of legal documents and he uses Jess in his work al Using this approach the original pattern might look like this lt isa gt lt variable gt x lt variable gt lt symbol gt mammal lt symbol gt lt isa gt This XML is much simpler but less general Each new deftemplate implies a new XML tag and in the case of an unordered deftemplate new tags for the slots too like this lt animal gt lt number of feet gt 4 lt
479. r the HVAC heating ventilation and air conditioning systems in a large building Jess will receive real time temperature data and based on these readings will send com mands to the multiple heating and cooling units in the building Although this is a specialized example the same principles would apply to controlling chemical plants intelligent traffic light sequencing monitoring automated assembly lines and even implementing manufacturing resource planning MRP systems In every case Jess has to receive data from the outside world and try to take action to affect the readings The Tax Forms Advisor from part 3 of this book and the PC Repair Assistant from part 4 only reasoned about facts in working memory Facts are jess Fact objects and they live entirely inside Jess When Jess is being used to directly react to events in the real world it only makes sense that Jess needs to be able to reason about objects outside of itself Section 6 5 talked about using the definstance function to add JavaBeans to working memory The application you re about to build uses this technique to interact with a collection of thermostats geothermal heat pumps and automated air vents Your first task and the topic of this chapter is to learn about the system you re going to control You will build a set of JavaBeans to represent and interact with the hardware and then begin to use them from Jess You ll also build a software simulator so that you can
480. ramework can create an instant application by storing the property values and connection information in a file and then later reconstituting the original Beans Generally Java s serialization API is used to do the pickling so most Java Beans implement the java io Serializable interface It s easy to add this capa bility to your own JavaBeans just declare that the class implements Serializable There are no methods in the Serializable interface it is a tagging interface used to signify to the Java virtual machine that it is OK to serialize the data from instances of the class Jess includes two built in functions bload and bsave that can save and restore Jess s state on any Java input and output streams using Java s serialization API If you intend to use these functions be sure any Java objects you add to work ing memory implement java io Serializable or these functions will fail You won t be using bload or bsave in the HVAC Controller application so it won t be necessary to make the Beans from this chapter implement this interface Summary In this chapter you began the work of using Jess to control the HVAC systems of an office building You defined a Java interface to represent Acme HVAC Systems C language library You wrote a simulator in Java to test your control algorithms against that interface You then wrote some JavaBeans which you can inject into Jess s working memory so that Jess can monitor the state of the
481. ranslate flowcharts into rules a Enhance the ask module from part 3 187 188 12 1 CHAPTER 12 Writing the PC Repair Assistant In this chapter you ll write a rule based system with a command line interface just like the Tax Forms Advisor You ll obtain the necessary knowledge by interviewing computer repair technicians develop the rules and test the system This applica tion reuses some techniques you developed for the Tax Forms Advisor For instance you ll modify and then reuse the ask module to ask the user for input You ll again use modules to represent processing phases The main knowledge base is in the MAIN module and we ll introduce a new control module named trigger Of course this application showcases some new techniques as well To model the technician s reasoning when diagnosing computer problems you ll use back ward chaining which we first discussed in section 7 4 Based on an initial descrip tion of the behavior of a computer the system will form a hypothesis about what might be wrong with it Backward chaining rules will then be triggered to either prove or disprove that hypothesis Along the way the program will make recom mendations for things the user might try to potentially fix the broken computer Many of the earliest rule based systems were backward chaining diagnostic pro grams like this one For example the MYCIN program discussed in section 2 2 1 diagnosed blood infections based on information ab
482. raserhead VHS EHVHS 28 969 9 Casablanca VHS CBVHS 1299 9 Repamanvhs fRmws fisso o Dumbo DVD Dovo fiss9 a The Matrix DVD Movo izs g9 9 TekMart C8 32 Video Cables mMCB32 2499 5 Bose Gold Standard Video Cables BGSVC 5499 apack Duracell AA Batteries DBAA4PK 399 a 10 pack TekMart AA Batteries TMAA10PK 1 99 9 Check Order Figure 19 3 The output of the catalog JSP 19 9 19 10 The Recommend servlet 343 Testing You can of course test a web application using a web browser but doing so can be both tedious and error prone Thankfully many automated testing tools are avail able for web sites These come in two often overlapping categories Load testing tools Simulate large numbers of simultaneous users visiting your site Page testing tools Invoke specific URLs and check that the resulting pages are correct The latter kind of testing is more interesting to developers at least in the early stages I like to use the open source HttpUnit ServletUnit framework for testing web applications http httpunit sourceforge net The HttpUnit part of the frame work lets you easily impersonate a browser and make requests to your web applica tion and then test the generated pages for correctness The ServletUnit part is a simple servlet container it lets you put a servlet through its paces directly and then test things like the values of the servlet s member variables afterward Both c
483. rderException NOTHING SingletonFuzzySet 0 3 SingletonFuzzySet 0 15 SingletonFuzzySet 0 06 SingletonFuzzySet 0 0 SingletonFuzzySet 0 06 SingletonFuzzySet 0 15 SingletonFuzzySet 0 3 outOfOrderEx catch XValueOutsideUODException outsideOUD NOTHING public Vent Hardware hw int floor m_hardware hw m_floor floor The rules for controlling the heat pump are the same as in the first fuzzy control example The rules to change the state of the vents now depend on the state of the heat pump the temperature on the floor and the rate of change of the tem perature You can build a pair of truth tables to show the correct vent changes for a temperature and rate of change Table 16 4 shows for the situation when the heat pump is cooling the table for the heating situation is similar Table 16 4 The recommended change to a vent for a given temperature and rate of temperature change Cold Cool OK Warm Hot Decreasing NB NM NS Z PS Zero NM NS Z PS PM Increasing NS Z PS PM PB The highlighted value NS in the middle row of the second column of table 16 4 represents this pseudo English rule Controlling with fuzzy rules 281 If the heat pump is cooling and the temperature is cool and the rate of change of temperature is zero Then adjust the vent position by a negative small amount In Jess you can write this rule like so
484. re satisfied by other purchases in the same order defrule remove satisfied recommendations If there are two products in the same category and one is part of an order and there is a recommendation of type requirement for the other part then remove the recommendation as the customer is already buying something in that category product part number part1 category category product part number part2 category category line item order number order part number part2 r lt recommend order number order part number part1 type requirement gt retract r You re finished writing the rules for the Recommendations Agent Recall that you re going to integrate these rules into a web application and that the web application will need access to the list of items in the current order the list of rec ommendations and the list of products in the catalog The web application can use queries to get access to this information In the next section you ll write some queries for the web application to call Some useful queries Queries let you efficiently access specified elements in Jess s working memory In this way they turn Jess into a relational database albeit a slightly strange one that doesn t use SQL When you build your web application you won t use a tradi tional database You ll use Jess to store all the application data instead Recall from section 7 7 that a defquery is like a
485. re stand ins for JavaBeans A shadow fact has a slot for every property of a JavaBean and for dynamic shadow facts defined using the definstance dynamic function those slots track the contents of the Java Bean s properties in real time Therefore everything about patterns in this chapter applies equally to pat terns that match facts and to patterns that match JavaBeans There s no way to tell by looking at a pattern whether it s intended to match deftemplate facts or shadow facts Constraining slot data The baby is wet fact in the previous section didn t have any slot data Most facts do however and most patterns need to specify some particular set of slot values Constraining slot data 101 for the facts they match These specifications are called constraints because they constrain the values a slot can have in a fact that matches the pattern A number of different kinds of constraints can be used to match slot data a Literal constraints Specify exact slot values a Variable constraints Bind a matched value to a variable m Connective constraints Let you combine conditions to match A and B or A orB a Predicate constraints Let you call a function to test for a match m Return value constraints Test for an exact match between a slot s contents and the result of a function call 7 2 1 Literal constraints Literal slot values can be included in patterns as constraints A pattern including a literal value matc
486. reate all the necessary JavaBeans and use definstance to connect the hard ware to Jess You need control Vent and control Thermometer objects for each floor and a control HeatPump for every three floors Because Jess doesn t have a built in function that works like Java s for loop you use Jess s while instead Tell Jess about the JavaBean classes defclass Thermometer control Thermometer 16 3 Controlling the heat pumps 259 defclass Vent control Vent defclass HeatPump control HeatPump Create the Vent and Thermometer Beans bind n n floors while gt n 0 do definstance Thermometer new control Thermometer hw n definstance Vent new control Vent hw n bind n n 1 Create the HeatPump Beans bind n n heatpumps while gt n 0 do definstance HeatPump new control HeatPump hw n bind n n 1 If you use the batch command to execute what you ve written so far you can then use the facts command to check that all the shadow facts exist If you check them repeatedly you ll see that the temperature data changes over time and stays in sync with the readings in the simulator GUI If you re using the Jess Win IDE http herzberg ca sandia gov jess user shtml you can watch the facts change in real time Controlling the heat pumps The heat pumps are controlled by a small cluster of rules You control the vents from a separate set of rules for clari
487. receive help from the author and from other users To access the forum and subscribe to it point your web browser to www manning com friedman hill This page provides information on how to get on the forum once you are reg istered what kind of help is available and the rules of conduct on the forum Manning s commitment to our readers is to provide a venue where a meaning ful dialog between individual readers and between readers and the author can take place It is not a commitment to any specific amount of participation on the part of the author whose contribution to the AO remains voluntary and unpaid We sug gest you try asking the author some challenging questions lest his interest stray The Author Online forum and the archives of previous discussions will be accessible from the publisher s web site as long as the book is in print XX X about the title By combining introductions overviews and how to examples the Jn Action books are designed to help learning and remembering According to research in cogni tive science the things people remember are things they discover during self motivated exploration Although no one at Manning is a cognitive scientist we are convinced that for learning to become permanent it must pass through stages of exploration play and interestingly retelling of what is being learned People understand and remember new things which is to say they master them only after actively explor in
488. red and unordered facts are useful in many situations but in many real world applications it s useful to have rules respond to things that happen outside of the rule engine Jess lets you put regular Java objects in working memory instances of your own classes that can serve as hooks into a larger software system as long as those objects fulfill the minimal requirements necessary to be JavaBeans As previously mentioned unordered facts look a bit like Java objects Specifi cally they look a lot like JavaBeans The JavaBean component architecture specifi cation http java sun com products javabeans defines a JavaBean simply as a self contained reusable component that can be used from a graphical builder tool JavaBeans like other kinds of software components for instance Visual Basic controls often serve as interfaces to more complex systems such as data bases or special hardware JavaBeans have slots The similarity between JavaBeans and unordered facts is that both have a list of slots for JavaBeans they re called properties containing values that might change over time There s plenty more to JavaBeans than just properties however those features go beyond the scope of this book A JavaBean property is most often a pair of methods named in a standard way If the property is a String named label the Java methods look like this String getLabel void setLabel String The get method is used to read the v
489. remember that Jess will be simultaneously processing many line items belonging to many orders placed by many different customers There s nothing wrong with this scenario and Jess will handle it very well How ever it will be important to make sure that all the rules identify the specific cus tomer and order they are processing that is you can t accidentally recommend products one customer might like to some second unrelated customer It will be easy to do this but it is important to keep in mind About testing The rules you re about to write will eventually be embedded in a web application When they are they won t have any command line or GUI access so it will be hard to debug them at that point Therefore it s important to test these rules as you write them You can put some product order and line item facts in a sepa rate file and load them using load facts when you need to use them as test cases Try to design test facts to probe each individual rule and think about what the correct results should be before you run a test For example imagine you ve written a rule like this one defrule recommend rubber duck customer customer id id not and order order number order customer id id line item order number order part number part product part number part name Rubber duck gt assert recommend order number order part number part This rule says If there s a custo
490. research 151 156 determinism 96 device drivers 325 diagnostic programs 188 dialog box 214 diff 430 Digital Equipment Corporation 292 DimmerSwitch 89 dispatch 334 distance 55 distributed 359 div 397 do backward chaining 117 118 397 Document Object Model See DOM documentation comment 56 97 doGet 328 337 338 DOM 373 domain 17 150 152 domain expert 155 256 Donne John 222 doPost 338 double negative 111 DTD 367 368 ducklessness 295 duplicate 85 94 107 397 duplicating facts 85 DVDs 293 E e 397 editability 361 efficiency 141 144 297 INDEX 435 EJBs 376 378 restrictions 377 electronic commerce 37 Emacs 46 jess mode 46 Emacs Lisp 58 email filtering 9 EMYCIN 19 encapsulation 352 engine 211 398 enterprise 358 enterprise application 358 Enterprise JavaBeans See EJBs enterprise systems 355 entity beans 376 environment variables setting 34 eq 398 eq 398 error checking 204 error handling 226 eval 55 398 evaluating expressions 51 function call 46 event adapter 208 event handlers 208 209 event object 208 event thread 209 exception handler 71 exceptions 164 catching 71 ignoring 317 throwing 73 executeCommand 308 338 efficiency 310 execution context 239 416 execution engine 19 23 exists 113 exit 232 398 exp 398 experimentation 205 expert knowledge 159 expert system shell 19 expert systems 18 37 expert users 155 explanation facility 184 436 INDEX explanations 181 expli
491. ressions that follows the sym bol then in order If the expression is FALSE then the statements after else are evaluated if the optional else clause is present This example uses the function member which accepts a value and a plain list as arguments and returns TRUE if the value appears in any position of the list Control flow 53 Jess gt bind grocery list create eggs milk bread eggs milk bread Jess gt if member S eggs grocery list then printout t I need to buy eggs crlf else printout t No eggs thanks crlf I need to buy eggs Either the if block or the else block can be terminated early using the return function The return value of the whole if function call is the value of the last expression evaluated You can chain if functions together much as you can in Java All you need to do is to use a second if function call as the body of an else block In this exam ple three if function calls are nested together Jess gt bind x 1 1 Jess gt if x 3 then printout t x is three crlf else if x 2 then printout t x is two crlf else if x 1 then printout t x is one crlf x is one The if function like the while function takes a single Boolean expression as its first argument What if you need to write several expressions to compute a Bool ean value The progn function is there to help 4 3 4 progn The progn function evaluates a list of expressions and retur
492. rief explanation of a single rule NOTE If you re interested in learning more about XSLT there are many excel lent books on the topic my favorite is by Michael Kay XSLT Programmer s Reference 2d ed Birmingham UK Wrox 2001 XSLT is a powerful technology and learning to use it helps you think declaratively a useful skill relevant to all rule based programming Here is an XSLT transformation rule from the script it s responsible for process ing variable nodes lt xsl template match variable gt lt xsl text gt lt xsl text gt 370 CHAPTER 20 Jess XML and the enterprise lt xsl value of select name gt lt xsl template gt An xsl template element specifies a rule The match attribute is a pattern that specifies where this rule applies this one applies to all variable elements in the XML xs1 text is an XSLT command that emits its contents verbatim When this script sees a variable element in the XML it prints a space and then a question mark which is how variable names in Jess begin By not including the question mark in the XML you make it easier to generate rules in other rule languages that don t use the question mark convention The xsl value of element emits a value derived from an element in the XML The select attribute dictates what the xs1 value of will do here it says to use the value of the name attribute of the current node the character indicates that the following string is an
493. rigger the firing of multiple other rules The modular and descriptive nature of individual rules makes a program like this easier to modify over time and easier to develop incrementally compared to a monolithic program without rules In the next chapter you ll again tinker with the ask module This time you ll give it a graphical interface and free the user from the need to type answers like yes and no at the command line Adding a graphical interface In this chapter you ll a Create a window a Fill it with graphical components a Write event handlers a all from the Jess language 203 204 13 1 CHAPTER 13 Adding a graphical interface These days virtually all computer applications come with a graphical user inter face GUI Graphical interfaces make many users feel more comfortable than command lines Furthermore by providing a bounded set of commands explicit in menus or implicit in radio buttons checkboxes and other controls GUIs pro vide guidance that helps users use the application efficiently and correctly One advantage of a GUI is perhaps less obvious Because it limits user actions it can make invalid user inputs impossible thereby simplifying the application s internal error checking code Your first real application the Tax Forms Advisor had only a command line interface The PC Repair Assistant you re developing now should have a friendlier GUI You ll develop such an interface in this chapter
494. right Each golfer wears different colored pants one is wearing red pants The golfer to Fred s immediate right is wearing blue pants m Joe is second in line Bob is wearing plaid pants Tom isn t in position one or four and he isn t wearing the hideous orange pants In what order will the four golfers tee off and what color are each golfer s pants You get the gist of it right away but how on earth are you supposed to figure it out There s no formula to use no analytic procedure for deriving a solution Algebra was one thing but this Why weren t you paying attention in class Rules to the rescue A rule based program can satisfy Mrs Rosencrantz by effi ciently finding the one combination of names positions and colors that fits all the constraints You can directly translate the problem statement into rules and the rules will find the solution Let s write that program to see how a rule based system would solve this prob lem You ll write the program in the Jess language Don t be concerned that you don t know the Jess language yet right now I d just like you to understand the approach You re going to 1 Choose a way to represent the possible combinations of men s names posi tions and pants colors 2 Write one rule that describes the problem Math class melee 5 The Jess rule engine will find the solution automatically Let s get started The first step is to define data structure
495. rk Rule engine standards Various commercial off the shelf products other than application servers can be designed to work together with rule engines Historically there has been a cer tain amount of vendor lock in because each rule engine has its own program mer s interface The Java Rule Engine API http www jcp org jsr detail 94 jsp defined by the javax rules package is a standard enterprise API for accessing rule engines currently being developed by a consortium of rule engine vendors under the Java Community Process The javax rules package will allow the host program to interact generically with multiple rule engines the same way the Java Database Connectivity J DBC API makes it possible to write vendor neutral database programs The API will include mechanisms for creating and managing sets of rules for adding removing and modifying objects in working memory and for initializing resetting and running the engine Soon perhaps Summary 27 even by the time you read this most popular rule engines including Jess will be accessible using the javax rules API In fact the official reference implemen tation of the javax rules API is currently slated to be a driver for Jess The javax rules API will not specify a standard rule language however Other groups are working on developing standardized rule languages although less consensus exists in this area For the same reason there is no one standard general progra
496. rmometerFV addTerm cold new RFuzzySet 68 5 69 new RightLinearFunction m_thermometerFV addTerm cool new TriangleFuzzySet 68 5 69 70 m_thermometerFV addTerm OK new TriangleFuzzySet 69 70 71 m_thermometerFV addTerm warm new TriangleFuzzySet 70 71 71 5 Vv Controlling with fuzzy rules 271 m_thermometerFV addTerm hot new LFuzzySet 71 71 5 new LeftLinearFunction catch XValuesOutOfOrderException outOfOrderEx NOTHING catch XValueOutsideUODException outsideOUD NOTHING public Thermometer Hardware hw int floor m_hardware hw m_floor floor new Thread this start public int getFloor return m_floor public double getReading return m_hardware getTemperature m_floor public FuzzyValue getFuzzyReading Fuzzy reading value return m_fuzzyReading public void run while true Send double reading getReading property boolean readingChanged reading m_oldReading change m_pcs firePropertyChange reading notification new Double m_oldRead ing new Double reading m_oldReading reading if readingChanged try m_fuzzyReading new FuzzyValue m_thermometerFV new SingletonFuzzySet reading catch XValuesOutOfOrderException xvorder System out println Error xvorder return catch XValueOutsideUODException xvuod System out println Error xvuod return
497. rsal travelers Although nothing is known of the designers engravers and workers who colored this illustration by hand the exactitude of their execution is evident in this drawing The Muger del Xeque is just one of many figures in this colorful col lection Their diversity speaks vividly of the uniqueness and individuality of the world s towns and regions just 200 years ago This was a time when the dress codes of two regions separated by a few dozen miles identified people uniquely as belonging to one or the other The collection brings to life a sense of isolation xxxi xxxii ABOUT THE COVER ILLUSTRATION and distance of that period and of every other historic period except our own hyperkinetic present Dress codes have changed since then and the diversity by region so rich at the time has faded away It is now often hard to tell the inhabitant of one continent from another Perhaps trying to view it optimistically we have traded a cultural and visual diversity for a more varied personal life Or a more varied and interest ing intellectual and technical life In spite of the current downturn we at Manning celebrate the inventiveness the initiative and yes the fun of the computer business with book covers based on the rich diversity of regional life of two centuries ago brought back to life by the pictures from this collection Part 1 Introducing rule based systems hat are rule based systems What ar
498. rther by the application of fuzzy logic The reality connection In this chapter you ll a Write a simulator in Java a Interface it to Jess using JavaBeans a Begin to write rules about JavaBeans 221 222 CHAPTER 14 The reality connection No man is an island entire of itself every man is a piece of the continent a part of the main John Donne These days very little software is written to stand alone Component architectures application servers and networking have made it easier than ever to assemble software out of disparate parts It s becoming increasingly rare to code a com pletely self contained application After all with so many high quality libraries components tools services and containers available why reinvent the wheel So far in this book you ve written two applications using Jess In each case you used Jess as a standalone system You didn t write or use a single line of Java code except Jess itself You ll occasionally write software this way but for the remainder of the book we ll look at more realistic systems that use Jess as a part of a larger whole Jess was designed to be embedded in other Java software You ve already seen how Jess can call Java functions now you ll see that outside Java code can just as easily call on Jess Each of the remaining applications you build will include a component written in Java In the next few chapters you ll develop the control software fo
499. rule without a right hand side It includes a set of patterns that match facts the same way a rule s patterns do The difference is that a rule is matched automatically whereas a query is triggered by calling the function run query This function returns a java util Iterator that represents the list of matches for the query s patterns The query all products is the simplest possible example it matches every product in the catalog defquery all products product The web application will be able to get a list of all the products in the catalog by calling the Jess function run query all products You ll need this to build an order form 302 17 7 1 CHAPTER 17 Jess on the Web Queries can also contain a variables declaration The variables declaration lists the arguments to the query The arguments you supply to the run query function are bound to the variables of the same names in the patterns so you can specify at runtime the specific values the query s patterns should match A query to list all the items in an order is a good example It should take one argument order which specifies the order number of interest defquery items for order declare variables order line item order number order part number part product part number part Each match returned by this query will contain an item number fact and the asso ciated product fact You d need this information to compute an order total print a sh
500. s When a global variable is created it is initialized to the given value When the reset command is subsequently issued the variable may be reset to this same value depending on the current setting of Jess s reset globals property You can use the set reset globals function to set this property An example will help Jess gt defglobal x 3 TRUE Jess gt x 3 Jess gt bind x 4 4 Jess gt x 4 Jess gt reset Jess will reset x to its initial value of 3 TRUE Jess gt x 3 Jess gt bind x 4 4 Jess gt set reset globals nil FALSE Jess gt reset This time x will not be changed TRUE Jess gt x 4 You can read about set reset globals and the accompanying get reset glo bals function in appendix A Multifields Multifields are another special kind of variable A multifield is a variable whose first characters are for example x Multifields have special meaning in only two contexts in the argument lists of deffunctions see section 4 4 and on the left hand sides of rules see chapter 7 They are otherwise equivalent to regular vari ables In fact the variable x and a multifield x refer to the same storage loca tion they re two names for the same thing Now that you understand Jess variables you re ready to learn more about the structure of Jess code in general We return now to our study of lists More about lists As you ve seen if you type a l
501. s MAIN initial fact Id is 0 1 Note that because pattern bindings have to refer to specific facts you must be careful when using them with some of the grouping conditional elements described in the following sections You can t use them with not or test condi tional elements for example and when using them with or and and conditional 108 7 3 7 3 1 CHAPTER 7 Writing rules in Jess elements you must be careful that the binding will apply to only one fact Jess lets you write ambiguous bindings but they may lead to errors at runtime depending on how the patterns are matched The next section presents some additional details on this issue Qualifying patterns with conditional elements We ve just been looking at increasingly sophisticated ways to match the data within individual facts Now we ll look at ways to express more complex relation ships between facts and to qualify the matches for entire facts Conditional elements CEs are pattern modifiers They can group patterns into logical structures and they can say something about the meaning of a match There s even one condi tional element test that doesn t involve matching a fact at all Before we begin let me caution you that many of these conditional elements have the same names as predicate functions we looked at in the last section There s an and conditional element and there s an and predicate function Although they may look similar they re e
502. s appendix lists every function included with Jess 6 1 In the argument lists number means any number and lexeme means any symbol or string boolean argu ments can be anything all values but the symbol FALSE are interpreted as true An argument type of expression means that anything is acceptable A fact argument means the actual text of a fact whereas a fact id is either a reference to a jess Fact object or the integer ID of a fact Square brackets denote optional arguments means one or more of the preceding argument means zero or more and means or Most of these functions are very flexible about the argu ments they ll accept Extra arguments are generally ignored and missing argu ments are defaulted in some cases Arguments that are filenames deserve special care A filename argument must follow Jess s rules for a valid symbol or string On UNIX systems this presents no particular problems but Win32 filenames may need special treatment In particu lar Pathnames should use either double backslash or forward slash instead of single backslash as directory separators and pathnames that include a colon or a space character must be enclosed in double quotes Some functions are actually constructs special forms built into the Jess lan guage and so don t appear in this list Constructs can t be called like regular func tions for example they can t appear on the right hand side RHS of
503. s in a nonnumeric argument Jess gt 1 2 3 Jess gt one two Jess reported an error in routine Value numericValue while executing one two 164 10 3 CHAPTER 10 Designing the application Message Not a number one type ATOM Program text one two at line 2 An empty slot in an unordered fact contains the value nil which is a symbol not a number If you write a rule that matches this empty slot and uses a mathematical function to do it an exception will be thrown during pattern matching like this Jess gt deftemplate number slot value TRUE Jess gt defrule print big numbers number value v amp gt v 10000 gt printout t v is a big number crlf TRUE Jess gt assert number Jess reported an error in routine Value numericValue while executing gt v 10000 while executing rule LHS TEQ while executing rule LHS TECT while executing assert MAIN number value nil Message Not a number nil type ATOM Program text assert number at line 13 If you plan to use mathematical functions on the left hand side LHS ofa rule it makes sense to add numeric defaults to any slots intended to hold numeric values The income and dependents slots of the user template will hold numbers so you should modify the template to look like this deftemplate user slot income default 0 slot dependents default 0 Now the income and dependents slots will
504. s language Armed with this knowledge you then progress through a series of fully developed applications chosen to expose you to practical rule based development The book shows you how you can add power and intelligence to your Java software What s Inside Introduction to rule based thinking a Jess language tutorial Complete examples of Tax forms advisor Diagnostic assistant Fuzzy logic controller Web agent J2EE apps Dr Friedman Hill is the developer of Jess A Principal Member of the Technical Staff at Sandia National Laboratories he lives in Gaithersburg MD ia MANNING 49 95 US 74 95 Canada FREEBIES s Binary version of Jess e Complete examples on the web For non commercial use a this book ae Ted Neward Author Server Based Java Programming the Jess book A nice balance between an introduction and a reference John D Mitchell Coauthor Making Sense of Java Friedman Hill writes clearly The topic is complicated and he does an excellent job explaining it I recommend this book Roedy Green Author The Java Glossary intuitive and clever examples that show the reader how to build intelligent Java applications with Jess Robert B Trelease Ph D UCLA Brain Research Institute www manning com friedman hill AUTHOR ONLINE Ebook edition available WAUSAU aiibi AUN wi eH 1 ence 49 8
505. s to make it work with fuzzy temperatures You must import the nrc fuzzy package store the fuzzy reading of the temperature define a fuzzy variable for the thermometer so you can add the terms cold cool OK warm and hot and notify Jess whenever the FuzzyValue for the temperature reading changes The code for this functionality is shown in listing 16 1 with the major changes highlighted in bold Because not every Thermometer object needs to define the FuzzyVariable you make it static Figure 16 2 shows the fuzzy sets that represent the terms for describing temperature 270 CHAPTER 16 Writing the rules cold OK hot 1 0 Figure 16 2 The fuzzy terms cold cool OK warm and hot and the possibility i 3 i A graphs of oo sac L F i 3 eer ___ temperatures they 0 0 68 0 69 0 70 0 71 0 72 0 represent Listing 16 1 Thermometer class with a new fuzzy reading property package fuzzy import nrc fuzzy public class Thermometer extends BeanSupport implements Runnable private Hardware m_hardware private int m_ floor i Create FuzzyVariable to private double m_oldReading represent temperature 1 private FuzzyValue m_fuzzyReading private static FuzzyVariable m thermometerFV null static try m_thermometerFV new FuzzyVariable thermometer 0 120 catch InvalidFuzzyVariableNameException nameEx NOTHING catch InvalidUODRangeException rangeEx NOTHING try m_the
506. s to represent the smallest useful pieces of a possible solution to the problem a link between a name and either a position or a color deftemplate pants color slot of slot is deftemplate position slot of slot is A deftemplate is a bit like a class declaration in Java While class objects have member variables deftemplates have slots Each slot is a placeholder for a spe cific piece of information For example the pants color template has a slot named of for a person s name and a slot named is to hold a color Whereas a Java class is a definition of a type of object a template is a definition for a type of fact a fact is basically what it sounds like a piece of possibly useful information A pants color fact represents the idea that one specific golfer named in the of slot has a certain color pants named in the is slot You ll use these templates to create facts representing each of the possible combinations There are 32 of them altogether for example pants color of Bob is red position of Joe is 3 You can write a rule to create all 32 of these facts and put them into working mem ory a kind of scratch space Jess uses to store the facts it knows defrule generate possibilities gt foreach name create Fred Joe Bob Tom foreach color create red blue plaid orange assert pants color of name is color foreach position create 1 2 3 4 assert position of name is positi
507. se and a code element which contains Jess code Note that the Jess code has to use normal XML encoding so that special characters like amp and gt must be rep resented as amp amp and amp 1t respectively Here s a rule base containing the single add two integers rule lt xml version 1 0 encoding UTF 8 gt lt rule execution set gt lt name gt Mathemetical rules lt name gt lt description gt One rule that adds two Integers description gt lt code gt defclass java lang Integer java lang Integer nil defrule add two integers 388 21 4 CHAPTER 21 Jess in the J2EE environment java lang Integer OBJECT 1i1 java lang Integer OBJECT i2 amp amp i1 amp gt undefinstance 1i1 undefinstance i2 bind i3 new Integer i1 intValue i2 intValue definstance java lang Integer i3 static lt code gt lt rule execution set gt The RI is bound by the limitations of the javax rules API As such there are no hooks to give you access to queries store fetch or other Jess features When the javax rules API is finalized Jess will offer native support for it including extensions to allow access to these features and more Summary This chapter has had a bit of a shotgun quality to it it hits a bit here and a bit there and misses little bits in between and there was no way to avoid writing it this way The J2EE specification is vast and there are many different ways y
508. sert item number name number TRUE Jess gt reset TRUE Jess gt assert do price check waffles lt Fact 1 gt Jess gt run Price of waffles is 1 99 3 You can wrap a special conditional element explicit around a pattern to inhibit backward chaining on an otherwise reactive pattern explicit can be used in any combination with all other conditional elements Most rule based systems consist of dozens if not hundreds of rules While such a Managing the agenda program is running at any one time a large number of rules may be simultaneously activated How does Jess decide which rule to fire next Read on to find out In section 7 1 you used watch activations and watch rules to observe the operation of a simple rule In particular you learned that a rule is activated when its 7 5 1 7 5 2 Managing the agenda 121 LHS matches working memory but it won t immediately fire The agenda is the list of rules that have been activated but haven t fired yet For some applications the agenda never contains more than one activated rule and so managing the agenda isn t a very interesting topic But in most applications the agenda contains multi ple rules at once When this is the case managing the agenda becomes important In this section we ll study how Jess chooses which rule to fire next from among all the activated rules on the agenda and how you can influence this choice Conflict resolutio
509. servlets three JavaServer Pages and one HTML file The resulting application is the largest one you ve developed so far and the first one you ve written that doesn t use jess Main Many details of a proper e commerce application are omitted but the embedded Recommendations Agent could form the core of an industrial strength online ordering web site In the next and final part of this book you ll see how rule engines in gen eral and Jess in particular can be used in enterprise applications Part 7 Enterprise systems usiness rules are policies and procedures by which corporations conduct their affairs Capturing business rules in a rule engine makes great sense This is but one of the many applications for rule engines in enterprise systems large scale mission critical software that supports business operations At the heart of many enterprise applications these days is the Java 2 Enterprise Edition J2EE environ ment J2EE is an enabling technology that combines access to data and comput ing resources into a standard service called an application server Part 7 of this book concerns the interaction between rule engines and application servers We ll touch on a wide range of relevant topics including using Jess with Enterprise Java Beans and with XML Jess XML and the enterprise In this chapter you ll Learn about rule engines in enterprise applications Look at XML representations for rules Transform Jess
510. set point gt set hp state p off This rule could be rendered in English as if there is a heat pump in the cooling state and there are no floors serviced by this pump with temperature above the set point then turn the heat pump off The moment of truth These four rules are enough to regulate the temperatures of the HVAC simulator Adding the vent control rules will refine the behavior but the basic heating and cooling control is already in place To make the program work you need to add calls to set vent state to open all the vents manually of course Testing the system at this stage shows that the rules defined so far suffice to turn the heat pumps on and off Without the vent control rules however the tem perature differential between floors has a tendency to force the system to over shoot When the system is in cooling mode some floors are cooled below the lower guard line while others are still in the comfort zone Unfortunately this set of rules turns the heat pump off and then immediately puts it into heating mode under these conditions You need to add rules to equalize the temperature between floors so this won t happen 262 CHAPTER 16 Writing the rules 16 4 Controlling the vents 16 4 1 Although this may seem obvious it s worth stating for the record the vents have two purposes a To let the hot or cold air in when needed To keep it out when it s not Therefore you need two d
511. set to 100 then the system again settles at a temperature close to 70 8 with the vents opened at about 0 7 With an outside temperature of 110 the system settles at a temperature near 71 3 This is quite good in that the temperature does not fluctuate much and stays within the guard temperatures However the application only works extremely well getting to 70 degrees and staying there for a small set of conditions The rules do not handle a widely varying set of initial conditions and the shape of the fuzzy sets should probably be changed to suit these situations In particular the definitions of hot cold and so on for temperature need to be adjusted for the current set point One option would be to modify the sets according to the new set point another would be to define the sets as a temperature difference from the set point This option is addressed in the next example It results in far less movement of the valves and less fluctuation of the temperature than the non fuzzy approach Version 2 relative vent positions Often control systems use not only the difference in the value from its set point but also the rate at which the change is taking place to determine how to adjust the control parameters In this case this behavior means monitoring not only the temperature but also the rate of change of the temperature over time In the first version of the fuzzy controller you defined the terms to describe the temperature based on a fixed t
512. sion gt Concatenates its arguments as strings to form a single symbol For Java objects the toString method is called It s easy to create unparseable symbols using this function For example although a Jess symbol normally can t contain a space the symbol returned by sym cat a space does contain a space If you use this symbol as the name of a rule for example and then save the rule s text into a file Jess will report a syntax error if you later try to read that file back in Therefore you should use sym cat carefully This advice also applies to the jess Value constructor that takes a string as an argument it can also be used to create invalid symbols system lt lexeme gt amp Sends a command to the operating system Each argument becomes one element of the argument array in a call to the Java java lang Runtime exec String 412 APPENDIX A Jess functions method therefore to execute the command edit myfile txt you should call system edit myfile txt not system edit myfile txt Normally blocks Jess stops until the launched application returns but if the last argument is an ampersand amp the program runs in the background The standard output and standard error streams of the process are connected to the t router but the input of the process is not connected to the terminal Returns the Java Process object You can call waitFor and then exitValue to get the exit status of the process throw
513. sions is supplied then those expressions are evaluated and the value of the last is returned implode lt list gt Creates a string from a list Converts each element of the list to a string and returns these strings concatenated with single intervening spaces import lt lexeme gt Works like the Java import statement You can import either a whole package using import java io ora single class using import java awt Button After a call to import all functions that can accept a Java class name new defclass call and so on refer to the import list to try to find the class that goes with a specific name Note that as in Java java lang is implicitly imported insert lt list gt lt integer gt lt expression gt Inserts one or more values in a list Inserts the elements of the second and later arguments so that they appear starting at the given one based index of the first list If any of the expressions are themselves lists the individual elements of these lists are inserted not the lists themselves instanceof lt object gt lt lexeme gt Returns TRUE if the object could legally be assigned to a Java variable whose class type is given by the second argument a Java class name Implemented using java lang Class isInstance The class name can be fully qualified or it can be an imported name see import 402 APPENDIX A Jess functions integer lt number gt Converts its only argument to an integer Trunc
514. situation perhaps an interview or an emergency In this case the rules are said to constitute an expert system Expert systems are widely used in many domains The newest applications of expert systems include being used as the reasoning part of intelligent agents in enterprise resource planning ERP systems and in order validation for elec tronic commerce The Jess language is also a general purpose programming language and it can directly access all Java classes and libraries For this reason Jess 1s also fre quently used as a dynamic scripting or rapid application development environ ment Whereas Java code generally must be compiled before it can be run Jess interprets code and executes it immediately upon being typed This allows you to experiment with Java APIs interactively and build up large programs incremen tally It is also easy to extend the Jess language with new commands written in Java or in Jess itself so the Jess language can be customized for specific applications Jess is therefore useful in a wide range of situations In this book you will see Jess used primarily for its rule engine capabilities but there will be plenty of scripting along the way How can you choose an architecture for a specific appli cation As with many things in Jess lots of choices are available Command line GUI or embedded Given its flexibility Jess can be used in command line applications GUI applica tions servlets and applets F
515. sm The rules in a module fire only when that module has the focus and only one module can be in focus at a time you ll learn about module focus in section 7 6 3 We ll discuss the following functions and constructs in this section m clear focus stack Empties the focus stack defmodule Defines a new module m focus Sets the focus module m get current module Returns the current module m get focus stack Returns the focus stack s contents as a list m list focus stack Displays the focus stack s contents pop focus Pops a module from the focus stack Defining constructs in modules So far in this book you haven t explicitly used modules If you don t specify a module by name when defining a rule or template it belongs by default to the cur rent module If you never explicitly define any modules the current module is always the main module which is named MAIN All the constructs you ve seen so far 124 7 6 2 CHAPTER 7 Writing rules in Jess have been defined in MAIN and therefore are often preceded by MAIN when dis played by Jess You can define a new module using the defmodule construct Jess gt defmodule WORK TRUE You can then place a deftemplate defrule or deffacts into a specific module by qualifying the name of the construct with the module name Jess gt deftemplate WORK job slot salary TRUE Jess gt list deftemplates WORK WORK job For a total of 1 deftemplates
516. son the user cares for Thinking about the general organization of the application suggests a few more possibilities question A question the system might ask the user answer An answer given by the user recommendation A note that the system will recommend a specific form These eight templates are good candidates for inclusion in the system Next you need to decide what form they will take ordered or unordered facts And for the unordered ones what slots should they have Filling in details Most facts in this system will represent physical or conceptual objects rather than commands or actions An object generally has observable properties color mass and so on To represent an object and its properties as a fact you can use an unor dered fact declaring an explicit deftemplate with multiple slots one for each property The user fact will clearly play a central role If you look back at the knowledge collected in section 9 3 1 you can see that the user s income and number of dependents are each fairly important and are each referenced in more than one place These two items are therefore good candidates to be slots in a user deftemplate which might look like this deftemplate user slot income slot dependents This is a good start but you need to worry about one detail default slot values Default slot values Jess s mathematical functions generally throw an exception to report the error if you pas
517. ss code is executed and applied to the given Rete object 418 APPENDIX B Abridged Java API for Jess public Value parse boolean prompt Does the parsing Returns the value of the last expression evaluated If prompt is true the Jess gt prompt is printed to the WSTDOUT router before each expression is read B 5 jess JessEvent If you request it using the setEventMask method in the Rete class Jess sends JessEvent objects as notifications when certain things happen Every JessEvent has a type that indicates what happened and an object that provides more informa tion usually it s the object most affected by the event The possible types their meanings and their associated objects are given in table B 1 Table B 1 Types of JessEvent Each type is a public static int variable in the JessEvent class Type Object type Description ACTIVATION jess Activation A defrule has been activated or deactivated CLEAR jess Rete A clear has been executed DEFCLASS The name of the defclass A defclass has been added or removed DEFFACT jess Deffacts A deffacts has been added or removed DEFGLOBAL jess Defglobal A defglobal has been added or removed DEFINSTANCE The JavaBean A definstance has been added or removed DEFRULE jess Defrule A defrule has been added or removed DEFRULE_FIRED jess Activation A defrule has been fired DEFTEMPLATE jess Deftemplate A deftemplate has be
518. ss is the best one but I hope you ll agree Each rule engine has its own strengths and weaknesses of course In general especially in the early stages of a large project it s good to avoid being locked into using one vendor s product If possi ble it makes sense to develop your rules in a vendor neutral language Unfortu nately no standard rule language is supported by all or even some of the major rule engine vendors In the Jess language rules are represented as defrule constructs Other rule engines have their own ways of representing rules including constructs in other programming languages text that s like natural language or graphical diagrams and flowcharts Each of these representations has advantages and disadvantages but it s important to remember that they re just that representations In general a core of common concepts can be expressed in all rule languages Although each language represents these concepts differently they all represent the same under lying information If rules that only use this common core of concepts are developed in a neutral flexible representation like XML then they can easily be translated into the native format supported by a specific rule engine as needed Of course this works only if the rule engine supports a textual representation for their rules If the only way to enter rules is through a GUI then the rule engine can t be used this way 362 20 2 2 20 2 3 CHAPTER
519. stantial difference is that before running the test this script builds the Java class test If the build fails the test fails too The file test java looks like this public class test public static void main String argv System out println Testing XXXXX TestMyFeature System out println Test done static void TestMyFeature System out println This is the test When you create a new Java based test from this template you need to replace XXXXX with a description of your test and implement the method TestMyFeature so that the output it prints demonstrates whether some feature of your code works correctly See the earlier discussion of the Jess template for ideas about creating the test ref file index Symbols 390 103 amp 104 390 30 391 390 as comment delimiter 44 lt 391 lt 100 lt 391 lt gt 391 107 391 gt 97 gt 391 gt 391 Pargv 58 ERROR 72 135 314 __data 87 101 313 104 297 implementation 106 104 A abs 391 abstract methods 225 accessors 66 Acme HVAC Systems 224 256 ActionEvent 208 ActionListener 208 actionPerformed 208 actions 17 activation 97 activation records 97 135 137 138 activation semaphore 422 addActionListener 208 addMultiSlot 313 addOutputRouter 329 addPropertyChangeListener 92 230 addSlot 312 addUserfunction 240 ad hoc algorithms 256 Adobe Photoshop 358 advice 58 60
520. sting the interview module is easy You can create a series of test cases each of which should contain the following A hypothetical tax situation The expected set of relevant questions The expected set of result facts Once you ve developed this set of test data run the program once by resetting setting the focus to interview and invoking run for each test case supplying the appropriate user answers for that hypothetical case Then check the results Make sure the system asks you all the questions on your list for that case and only those questions Use the facts command you ll need to use facts to see the facts in all modules to make sure all the expected facts are present and cor rect If you can capture your whole terminal session in a file by using the script command on UNIX for example then you can save the file and use it as part of an automated test script The program should produce the same output each time it is run you can use your automated script to check that changes you make later haven t broken the behavior of this module NOTE Many books have been written on the subject of software testing and al though such books are often useful they can be discouraging to read They generally concentrate on teaching you to find spanning sets of tests or lists of tests that cover every possible path through your application For most real applications of course such a spanning set is impossibly large The t
521. stion ident text Did you estion ident text Did you estion ident text text have dependent care expenses moving type yes no move for job related reasons employee type yes no have unreimbursed employee expenses reimbursed type yes no have reimbursed employee expenses casualty type yes no have losses from a theft or an accident on time type yes no too Will you be able to file on time Asking the user questions 175 question ident charity type yes no text Did you give more than 500 in property to charity question ident home office type yes no text Did you work in a home office 11 2 1 Income and dependents The user s income determines whether she can use either of the short forms so the system should ask about income first If the income is below 50 000 the num ber of dependents and the amount of interest income determine which of the short forms can be used Although the number of dependents is needed for other purposes the interest income is relevant only if the user s income is less than 50 000 therefore you make asking about interest income conditional on the user s income being below this limit To ask a question you assert an ask fact that indicates the question to be asked the ask module you developed in section 10 6 2 asks the question and asserts an answer fact The rules to ask these ques tions are shown in listing 11 2
522. stions the hardware diagnostic system can ask deffacts MAIN question data question ident hardware type multi valid x86 Macintosh other text What kind of hardware is it question ident sound type multi valid yes no text Does the computer make any sound question ident plugged in type multi valid yes no text Is the computer plugged in question ident seek type multi valid yes no 12 6 Testing 201 text Does the disk make seeking sounds question ident does beep type multi valid yes no text Does the computer beep question ident how many beeps type number valid text How many times does it beep question ident loose ram type multi valid yes no text Are any of the memory modules loose question ident boot begins type multi valid yes no text Does the computer begin to boot You can now run the complete application by calling reset and run until halt Testing I haven t mentioned testing so far in this chapter but don t take this to mean I ve forgotten about the test first strategy I talked about in previous parts of this book The larger and more complex the system the more important it is to test each part in isolation When I modified the ask module for example I reran all the tests for it that I developed in part 3 and I wrote several more I confirmed that ask would only accept the allowed answers for multip
523. structors 318 incorrect usage 317 root cause 317 throwing 318 toString method 317 jess version number 402 jess version string 402 JessWin 35 259 JFrame 205 setDefaultCloseOption 215 JNDI 378 384 JNI 224 join nodes 138 143 144 JOptionPane 214 JPanel 214 JScrollPane 206 jsp useBean 334 JSPs 291 295 332 341 377 forwarding servlet requests to 333 JSR 94 26 383 reference implementation 384 JTextArea 206 JTextField 207 JUnit 428 JVM support 289 JWSDP 291 K Kay Michael 369 keyboard 197 kill Userfunction 239 kiosk 150 kitchen sink 114 knowledge limits of 198 knowledge engineering 23 26 150 162 188 190 256 292 dressing the part 154 iterative process 151 overlapping development 153 knowledge engineers background 151 L Lao Tzu 150 late binding 57 layout managers 62 Leff Laurence 366 left memory 138 left hand side 17 97 length 106 402 lexeme 390 lexemep 402 LHS See left hand side libraries 156 224 library example 134 linguistic terms 273 Linux 324 326 Lisp 32 LIST 313 422 list deftemplates 124 402 list focus stack 127 402 list function 60 402 listp 404 lists 45 creating 49 nesting 50 plain 50 load facts 295 338 402 load function 239 247 403 load package 248 253 258 403 load testing 343 log 403 log10 403 logical 115 logical dependency 115 logical support 116 LONG 422 long 47 403 entering literal value 47 longp 403 loops 50 18
524. sub string 411 subtlety 265 Swing 188 204 208 symbolp 411 symbols 42 case sensitivity 43 sym cat 196 298 411 syntax diagrams 42 system 411 System in 319 System out 319 T t 319 I O router 319 Tabet Said 363 tagging interface 235 tags 139 Takagi Sugeno Kang 273 Tax Forms Advisor xxvii 147 150 157 162 173 192 taxable income 175 TekMart 335 temperature calculation 227 templates plastic 189 terminal nodes 138 144 test harness 169 test clp 429 test ref 429 test driven development 25 testing 24 167 169 173 177 182 183 201 215 253 261 263 295 343 428 coverage 177 201 framework 254 GUIs 215 in isolation 201 tests 113 automated 25 automated framework 25 conditional element 98 developing 25 textbooks 156 thermostats 282 thin client 290 292 thing factory 408 thread safety 308 threads 209 227 264 throw 412 Jess function 73 Throwable 316 time 412 toArray 67 tokens 42 139 deleting 139 number 43 string 44 Tomcat 285 324 326 and JSPs 333 default port 329 top level window 205 toString 63 transactions 358 TRUE 43 truth tables 256 259 262 280 try 197 412 Jess function 71 two input nodes 137 U unconditional support 116 undefadvice 59 412 undefinstance 315 413 undefrule 97 111 413 union 413 UNIX 169 177 428 unordered facts examples See facts unor dered unreimbursed expenses 176 untyped 47 unwatch 77 413 upcase 413 use ez form 152 user sessio
525. t questions You can fix this problem by telling the backward chaining mechanism not to try to trigger the right architecture rule and then asserting the appro priate MAIN ask fact at the right time You can modify the right architecture rule using the explicit conditional element which cancels backward chaining requests for a single pattern defrule MAIN right architecture declare auto focus TRUE explicit answer ident hardware text x86 gt recommend action consult a human expert halt Adding a MAIN ask hardware fact at the end of deffacts question data makes the hardware question come out first Give the application a try and make sure the conclusions it reaches are the same that you d arrive at by following the flowcharts Summary In this chapter you saw how backward chaining can be used to do something automatically on demand in this case you used it to schedule questions to be asked using a single generic rule The combination of forward chaining infer ence rules and backward chaining infrastructure rules is a powerful and common pattern in Jess systems Another common application of this idea is fetching needed facts automatically from a database Flowcharts are a reasonable way to represent some kinds of knowledge and you used them to good effect in this chapter The collection of inference rules in this chapter was more complex than you ve seen before most rules that fire can t
526. t defrule large order and no dairy shopping cart customer id id contents c amp and gt length c 50 not or member milk c member butter c checking out now id gt printout t Don t you need dairy products crlf TRUE Note that internally Jess implements the connective by transforming the whole pattern for that slot into predicate functions and then using or to represent the When evaluating a predicate constraint Jess interprets any return value except FALSE as if it were TRUE The member function returns FALSE if the given value is not a member of the list argument otherwise it returns the position of the value in the list Even though members never returns TRUE it works perfectly well as a predicate function because the non FALSE values are interpreted as TRUE Return value constraints Often you ll want to constrain the contents of slot to match the return value of a function For example if you wanted to find a pair of grocery items such that the price of one was exactly twice the price of another you might use a predicate con straint like this item price x item price y amp eq Py x 2 The eg function returns TRUE if the arguments are all equal or FALSE otherwise Although this approach works it s not especially pretty It would be more conve nient to write this using a return value constraint A return value constraint includes 7 2 6 Constrain
527. t facts f 0 MAIN initial fact f 1 MAIN product 354 sticky notes S1 99 f 2 MAIN product 355 paper clips S0 99 f 3 MAIN product 356 blue pens 2 99 f 4 MAIN product 357 index cards S0 99 f 5 MAIN product 358 stapler S5 99 For a total of 6 facts 6 2 Just the facts ma am 81 Whenever reset is called the facts in the deffacts are asserted This is a conve nient way to set up the product catalog in working memory for the first time and to restore it after working memory has been cleared Using the functions we ve discussed in this section you add and remove facts from Jess s working memory and see the effects of your changes Now let s exam ine the individual facts themselves in more detail Just the facts ma am Jess s working memory is stored in a complex data structure with multiple indexes so that searching the working memory is a very fast operation The working memory is therefore something like a relational database and the individual facts are like rows in a database table see figure 6 2 This implies that facts like table rows must have a specific structure so that they can be indexed Jess offers three different kinds of facts each kind has its own structure and is indexed and used in a different way Unordered facts are the workhorses of the working memory An unordered fact is quite literally like a row in a relational database table with individual named data f
528. t In this code you pass a Jess list made by create Jess automatically converts it to the needed array With the JComboBox defined you can modify ask user one final time deffunction ask user question type valid Set up the GUI to ask a question qfield setText question apanel removeAll if eq type multi then apanel add acombo apanel add acombo ok acombo removeAlliItems foreach item valid acombo addItem item else apanel add afield apanel add afield ok afield setText frame validate frame repaint This version is somewhat simpler because it doesn t have to compose a complex prompt string Instead based on the question type it installs one of two possible sets of components in the JPanel at the bottom of the window Figure 13 6 shows what the application looks like with the combo box panel installed Finishing touches The current version of the recommend action function still prints its recommen dations to the command line Java offers the convenient JOptionPane class as a quick way to post a modal dialog box You can easily modify recommend action to use it 13 7 Testing the interface 215 deffunction recommend action action Give final instructions to the user call JOptionPane showMessageDialog frame str cat I recommend that you action Recommendation get member JOptionPane INFORMATION_MESSAGE The argume
529. t pump states It uses the getSimulator method from InitSimulator the next few classes will use this method too The Userfunction for setting the heat pump state is similar but it must do the mapping between state names and numeric states in the reverse order package control import jess class WriteHeatPump implements Userfunction public String getName return set hp state public Value call ValueVector vv Context c throws JessException int heatPump vv get 1 intValue c Simulator sim InitSimulator getSimulator c String szState vv get 2 stringValue c 252 CHAPTER 15 Extending the Jess language try int state State heatpump szState sim setHeatPumpState heatPump state catch IllegalArgumentException iae throw new JessException set hp state Invalid state szState return Funcall TRU GI Finally you need two more Userfunctions to read and write the state of the vents These are similar to the last two but because the vent state is a Boolean value they are a bit simpler the function to read a vent s state can dispense with error handling altogether package control import jess class ReadVent implements Userfunction public String getName return get vent state public Value call ValueVector vv Context c throws JessException int vent vv get 1 intValue c Simulator sim InitSimulator getSimulator c boolean state sim getV
530. t conditional element The not conditional element matches the absence of a fact Its implementation involves a special kind of join node and a special field in every token The special field is called the negation count The special NOT join node uses the negation count to keep track of the number of times the node s conditions are met for each token in the left memory The count is incremented whenever the node s tests pass for a given token and decremented whenever they do not When the count reaches zero for a given left memory token a new token is formed from it and a special null fact which represents the nonexistent right memory input and sent to the node s output The important thing to notice about this description is that although the facts matching the pattern in the not conditional element arrive at the NOT node s right input only the tokens arriving at the left input form part of the output tokens A moment s reflection shows that it has to be this way the not conditional element is matched when there is no matching fact If there are no tokens in the right memory that means the not conditional element will always succeed so the NOT node will pass every token received on the left input directly to its output Now what should a NOT node do if it doesn t have a left input Such a case would arise if a not conditional element was the first pattern on the LHS of a rule or the first pattern in a given branch of an or condition
531. t expression random Returns a pseudo random integer between 0 and 65536 read lt lemexe gt Reads a single symbol string or number from the named router and returns this value The router t is the default if no argument is given By default newlines are treated as ordinary whitespace If you need to parse text line by line use readline and explodes readline lt lexeme gt Reads an entire line as a string from the specified I O router If no argument is given the default is t replace lt list gt lt integer gt lt integer gt lt list gt Replaces the specified range of a list value with a set of values The variable number of final arguments is inserted into the first list replacing elements between the one based indices given by the two numeric arguments inclusive Here s an example Jess gt replaces create a b c 2 2 create x y z a x y Zc reset Removes all facts from working memory and all activations from the agenda asserts the fact initial fact asserts all facts found in deffacts constructs asserts a fact representing each registered definstance and if the set reset globals property is TRUE initializes all defglobals rest lt list gt Returns all but the first field of a list as a new list like cdr in traditional LISP retract lt fact id gt Removes the given facts from working memory Retracting a shadow fact results in an implicit call to undefinstance for the correspondi
532. t make up the Recommendations Agent application Source file Description index html Simple login screen Catalog java Servlet to create the order object and list products catalog jsp JSP to display the product table and a form for ordering Recommend java Serviet that accepts the submitted order form and runs the Recommendations Agent recommend jsp JSP to display recommendations and a form for adding them to the order Purchase java Servlet to accept the completed order purchase jsp Final screen We ll concentrate on the Java code in both the servlets and the Java parts of the JSPs The HTML in the JSPs is simple minimalist and maybe even ugly This is typical The programmers produce stub JSPs with minimal HTML content and the web designers then flesh them out modifying the HTML parts and leaving the Java parts alone The index html login screen and the purchase servlet and JSP are fairly trivial in this implementation you won t collect credit card and shipping information The catalog and Recommend servlets and JSPs are the interesting parts and now finally you can get started The login screen The first page the customer sees is a login screen A real login screen might col lect a password as well as a username but you ll leave that aside for now All you need for the login screen is a simple HTML form that lets the customer enter a single string lt HTML gt lt HEAD gt lt TITLE gt Welcom
533. t might be too cold As soon as the pump was switched off the state of the system would be indistinguishable from the case where the building needed to be heated so the heating rules would switch the heat pump into heating mode Switching back and forth between heating and cooling mode is bad for a heat pump so this alterna tive is also unacceptable 16 3 2 Controlling the heat pumps 261 The second case running until the temperature dips below the set point leads to gradual switching because a range of temperatures is allowed The average temperature is then above the set point most of the time the room is warmer than the set temperature but not by much Most importantly though the heat pump cycles between cooling mode and off mode without any danger of being switched into heating mode which is exactly what you want Therefore your rule will use the actual set point as the shut off criterion If the heat pump is in cooling mode and the temperature of all connected floors is below the set point then the heat pump will be shut off This behavior corresponds to what is shown in the Cooling row in table 16 2 A similar argument would let you substitute heating and above for cooling and below Again you ll write this as two rules and PII just show the cooling version here defrule floors cool enough HeatPump state cooling number p not Thermometer floor f amp p which pump f reading t amp gt t
534. t the kinds of rules you write for that engine either by limit ing the possibilities or by encouraging you to write rules that would be particu larly efficient The agenda Once your inference engine figures out which rules should be fired it still must decide which rule to fire first The list of rules that could potentially fire is stored on the agenda The agenda is responsible for using the conflict strategy to decide which of the rules out of all those that apply have the highest priority and should be fired first Again this is potentially a hard problem and each rule engine has its own approach Commonly the conflict strategy might take into account the specificity or complexity of each rule and the relative age of the pre mises in the working memory Rules may also have specific priorities attached to them so that certain rules are more important and always fire first As an example the driving robot s control program might have two rules like these IF the light is green THEN go END 2 3 6 2 4 2 4 1 Developing rule based systems 23 IF a person is in front of you THEN stop END If the robot is stopped for a red light and the light turns green when someone is still in the crosswalk then both rules will apply It is important that the second rule fire before the first or the future of driving robots will be in serious peril This second rule should therefore be given a very high priority The exec
535. tant principle One Java class can implement multiple Userfunctions A call method can choose from among several possible actions depending on the name the function is invoked with If several Userfunction implementations need to share code as the HVAC functions surely will packaging them into a single class this way makes sense Some of Jess s built in functions use this technique Userfunctions and serialization As we discussed in section 14 4 Jess s bload and bsave functions use Java s serial ization API to preserve and restore Jess s state from any storage medium a file a network pipe and so forth More formally these functions save or restore all the member data of the active jess Rete object Therefore all the Java objects that are part of the Rete object s state should implement the java io Serializable interface This is a tagging interface which means it contains no methods and just serves as a signal to the Java virtual machine that serializing objects of the given class is OK Because the Userfunctions registered with Jess become part of a Rete object s state it is generally important that all your Userfunction classes imple ment this interface unless you re sure you ll never use bload and bsave when they are registered Because the current program will be controlling hardware which obviously can t be stored in a file it won t be necessary for the Userfunction classes to implement Serializable either
536. tant that these rules be auto focus rules in their own module Because they delete individual recommendations these rules could oth erwise thrash back and forth with the recommendation rules Each time a recommend fact was deleted one of the recommendation rules could acti vate and fire putting the same fact back into working memory only to have it be deleted again in an endless chain reaction This kind of prob lem is sometimes called an assertion storm You can stop an assertion storm either by removing all the requisite facts or by calling the Rete halt method which forces that Rete object to halt immediately Summary In this chapter you ve developed a basic set of rules for a Recommendations Agent It s important to realize that you could add many other rules to this system without changing any of the data structures With minor changes you could add rules to support sale prices and two for one deals However this small core of rules is enough to support the goal here which is to see how Jess can be embedded in web applications Summary 305 You ve also seen your first practical examples of using the defquery construct You wrote several queries which the web application implementation code will need to use In chapter 19 you ll build a web application around the Recommendations Agent rule base Before you can do that however you need to learn about using Jess from Java code This is the subject we ll tackle in
537. te object the function is executing in as well as the values of relevant variables public Rete getEngine Returns the Rete object relevant to the calling code public Value getVariable String name Returns the value of a variable The parameter is the name of the variable with no leading or characters You can use this method to get the current value of a defglobal public void setVariable String name Value value Sets the value of a variable The first parameter is the name of the variable with no leading or characters jess Fact The Fact class represents one entry in Jess s working memory Fact is a subclass of ValueVector public Fact String name Rete engine Asks the Rete object for a template by the given name if one doesn t exist an ordered one is created This is the most commonly used constructor public Deftemplate getDeftemplate Returns the Deftemplate for a Fact which can tell you the name of the Fact the number of slots and the names of its slots public int getFactld Returns a Fact s numeric identifier every Fact has one jess Jesp 417 public int getShadowMode If this Fact is not a shadow fact returns the constant Fact NO Otherwise returns Fact DYNAMIC or Fact STATIC depending on the type of shadow fact it is public Value getSlotValue String name Returns the value held by the named slot Ordered facts have a single slot named __data containing a list of items pu
538. ted out that this popular proverb is properly called Finagle s Law and that the orig inal formulation of Murpny s Law was If there are two or more ways to do something and one of those ways can result in a catastrophe then someone will do it I chose not to go against popular us age here but the pedant in me appreciated this fact enough to add a footnote For more information see The Jargon File for instance http info astrian net jargon terms f Finagle_s_Law html 18 CHAPTER 2 What are rule based systems of computer programs These programs are sometimes called rule engines A rule engine doesn t contain any rules until they are programmed in A rule engine knows how to follow rules without containing any specific knowledge itself A rule engine is generally part of a rule development and deployment environment The features offered by these environments vary widely depending on the intended applications for the rule engine and on the type of programmer intended to develop the systems This book will show you how to develop and deploy rule based systems in general To do so it will use the Jess rule engine in all its examples 2 2 1 Expert systems Expert systems rule based computer programs that capture the knowledge of human experts in their own fields of expertise were a success story for artificial intelligence research in the 1970s and 1980s Early successful expert systems were built around rules sometimes cal
539. ted to show a fact being removed In the rest of this chapter we ll show Jess s output as if the watch function has not been issued but you can use it whenever you want to keep an eye on what s happening with working memory As you may have inferred from the example the unwatch function reverses the effect of watch The facts function Although watch can tell you when new facts appear and old ones are removed it can t give you the big picture You can get that using facts You can see a list of all the facts in the working memory using the facts function If you continue the previous example by typing facts at the prompt you ll see that the fact MAIN initial fact is the only one in working memory Jess gt facts f 0 MAIN initial fact For a total of 1 facts The initial fact turns out to be useful but it s not especially interesting Let s add some more facts to working memory so you have something worth looking at 78 CHAPTER 6 Representing facts in Jess Creating facts with assert Rules can only act on information that is represented by facts in Jess s working memory In any nontrivial program then you need to create new facts of your own New facts are added to the working memory using the assert function Jess gt reset TRUE Jess gt assert groceries milk eggs bread lt Fact 1 gt Jess gt facts f 0 MAIN initial fact f 1 MAIN groceries milk eggs bread For a total of 2
540. tegy interface and then specifying its fully qualified class name as the argument to set strategy show deffacts Displays all defined deffacts to the t router show deftemplates Displays all defined deftemplates to the t router show jess listeners Displays all JessListeners registered with the engine to the t router socket lt lexeme gt lt integer gt lt lexeme gt Somewhat equivalent to open except that instead of opening a file opens an unbuffered TCP network connection to the host named by the first argument at the numbered port given by the second argument and installs it as a pair of read and write routers under the name given by the last argument Because Jess routers can t work with binary data this method is mostly useful to let instances of Jess communicate over a network For general networking it would make sense to use the java net package directly sqrt lt number gt Returns the square root of its argument store lt lexeme gt lt expression gt Associates the expression with the name given by the first argument such that later calls to fetch using that same name will retrieve it Storing the symbol nil clears any value associated with the name Analogous to the store member function of the jess Rete class The storage area is local to the Rete engine where store is called str cat lt expression gt Concatenates its arguments as strings to form a single string without intervening spaces
541. tension adds a multislot to the question template to hold the possible values for multiple choice questions The ask question by id rule needs to match this multifield and pass it along to an extended version of the ask user function ask user in turn needs to pass the multifield to an enhanced is of type function which can validate questions of type multi numeric and text The new template and rule are shown in listing 12 4 and the modified functions appear in listing 12 5 deftemplate question A question the application may ask slot text The question itself slot type Can be multi text or numeric multislot valid The allowed answers for type multi slot ident The name of the question defrule ask ask question by id Ask a question and assert the answer declare auto focus TRUE MAIN question ident id text text valid valid type type not MAIN answer ident id ask lt MAIN ask id gt retract ask bind answer ask user text type valid assert answer ident id text answer return 196 CHAPTER 12 Writing the PC Repair Assistant Listing 12 5 Modifications to the question asking functions deffunction ask user question type valid Ask a question and return the answer bind answer while not is of type Panswer type valid do printout t question if eq type multi then printout t crlf Valid answers ar
542. tential compatibility problems Powerful servers may be needed to meet demand Because the configuration and maintenance issues are much simpler thin client architectures are usually the preferred way to deploy an application on the Web In chapter 19 you will develop the Recommendations Agent using a thin client architecture 17 2 A Jess application for the Web 17 3 One of the first and most famous rule based applications was Digital Equipment Corporation s XCON order configuration system see section 1 2 2 XCON helped sales consultants configure orders for DEC mainframe computers making sure the order included all the accessories the customer needed The grandchildren of XCON are now on display all over the Web Every web site that helps you select compatible options when you buy a computer online owes an obvious tip of the hat to XCON Not quite so obvious perhaps are sites that recommend additional purchases the way a good sales clerk might This top would look cute with those pants Rule based order configuration and recommendation systems are common today In this chapter and the next two you ll build a system that combines both of these tasks It will recommend items the customer probably needs as well as items it thinks the customer might just enjoy Knowledge engineering Yow re going to write a fairly general system to look at what a customer is ordering and suggest other things they might want to buy
543. ter 6 but for now we ll only be concerned with setters and getters Many of the Java library classes especially in the graphical libraries use this method naming convention 1 The JavaBeans project page is at http java sun com products javabeans it offers the JavaBeans spec ification for download 5 2 4 Calling Java methods 67 Jess includes the functions set and get which can be used as an alternative to call for setters and getters The following pairs of function calls are equivalent Jess gt bind b new javax swing JButton lt External Address javax swing JButton gt Jess gt b setText Press Me or Jess gt set b text Press Me Jess gt b getText or Press Me Jess gt get b text Press Me The name of a setter or getter method includes a property name which is text in these examples The property name is passed as the second argument to the set or get functions To derive the property name to use remove the prefix from the Java method name and make the initial capital letter of the rest of the name low ercase The one exception is for names like getURL where the property name is URL in all uppercase This convention is the same as that used by the JavaBeans specification So far we ve dealt only with single Java objects and values Jess also lets you work with Java arrays Working with arrays The grocery price table can also serve as a simple grocery list You can ask a J
544. terface and then calling set strategy with the name of your class as the argu ment The Strategy interface has a single nontrivial method compare that com pares two activations and returns 1 1 or 0 to signify that the first activation the second activation or either one should fire first The conflict resolution strategy determines how activations are ordered based on when they are added to the agenda Sometimes though you may find that you want to fine tune the ordering a bit You can use salience to accomplish this Changing rule priority with salience Sometimes you may find that a particular rule should be treated as a special case by the conflict resolution strategy A rule that reports a security breach might 122 CHAPTER 7 Writing rules in Jess need to fire immediately regardless of what else is on the agenda On the other hand a rule that cleans up unused facts might only need to run during the idle time when no other rules are activated You can tell the conflict resolver to treat these rules specially using rule salience Each rule has a property called salience that acts as a priority setting for that rule Activated rules of the highest salience always fire first followed by rules of lower salience Within a set of rules with identical salience the order is deter mined as described in the previous section You can use a salience declaration to set the salience of a rule Jess gt defrule defer exit until agenda empty
545. test ref JavaTemplate Makefile README runtest sh runtest ref test java test ref The framework works as follows The top level runtest sh finds all the runtest sh files in subdirectories and executes them one at a time Each runtest sh in a subdi rectory runs its own test You can edit the script in a particular subdirectory to do whatever you want but the default scripts run a program capture the output in a file and compare that to a reference file If the files match Test succeeded is dis played if they don t a failure message shows up instead The top level runtest sh skips any subdirectory in which it finds a file named runtest sh disable this lets you turn off tests if necessary There is a README file in each template directory It s a good idea to use it to write down a short description of what s being tested whenever you create a new test In the text as we developed the tutorial applications throughout this book I described various tests you could do and showed example dialogs confirming that functions or rules behaved as expected If I were developing these systems by myself Pd make each of those examples into an individual test Then each time I added a new test I d run all the existing tests to be sure everything still worked You can never have too many tests Now let s look at the two test templates The Jess template The two test templates are similar The Jess template s
546. th a certain grade or degree or possibility of membership This grade of membership in the fuzzy set is usually represented by values from 0 to 1 with 0 meaning definitely not a mem ber of the set and 1 meaning definitely is a member of the set For example the fuzzy term young might be defined and shown graphically as in figure 16 1 There you can see that between 0 and 25 years of age the membership value is 1 defi nitely young above 50 it is 0 definitely not young and between 25 and 50 the membership value decreases partial membership in the set 1 0 u young 10 20 30 40 50 60 70 80 Figure 16 1 Possibility distribution of young Controlling with fuzzy rules 267 Representing fuzziness in FuzzyJ FuzzyVariable FuzzySet and FuzzyValue Fuzzy concepts are represented using fuzzy variables fuzzy sets and fuzzy values in the Fuzzy Toolkit A FuzzyVariable defines the basic components used to describe a fuzzy concept It consists of a name for the variable for example air temperature the units of the variable if required for example degrees C a range of valid values from 0 to 100 and a set of fuzzy terms that can be used to describe the particular fuzzy concepts for this variable The fuzzy terms are described using a term name such as cold along with a FuzzySet representing that term A FuzzyValue is normally created by specifying a FuzzyVariable and a fuzzy expression The following example shows how this
547. that it is a self contained logical system Because the relationships are built into the lan guage any software that understands DAML can make inferences based on the data without any additional information Although DAML currently doesn t include a specification for representing rules one is in the works As with RuleML people have written tools that translate between DAML and Jess In particular Mike Dean has several examples on the DAML web site that demonstrate conversion of DAML into Jess facts XML based rule representations 365 20 3 3 Homegrown representations 20 3 4 The RuleML and DAML projects are promising but they re also still being actively developed Tools for working with either of these complex standards are still evolv ing and it will probably be some time before they come into widespread use One of the more serious limitations of existing standard rule languages is their minimal support for describing actions on the right hand side RHS of a rule As you ve seen in many practical Jess applications the RHSs of rules can be com plex and working within the confines of a standard representation can be limit ing The situation will doubtless improve in the future but for now this is a significant problem Currently most people using XML and rules together use a homegrown repre sentation of their own devising If you come up with your own way of describing rules in XML it s likely to be simpler than either
548. the Recommendations Agent rules you developed in chapter 17 18 1 Getting started with the Jess library Let s begin at the beginning The core of the Jess library is the jess Rete class see section 13 4 1 An instance of jess Rete is in a sense an instance of Jess Every jess Rete object has its own independent working memory its own list of rules and its own set of functions The Rete class exports methods for adding finding and removing facts rules functions and other constructs The Rete class is a facade for the Jess library Although there are many other classes in the library jess Rete provides a convenient central access point It s easy to create a jess Rete object There is a default constructor one that accepts no arguments import jess Rete engine new Rete The most commonly used constructor however accepts a java lang Object argument The Rete object stores this Object and whenever it needs to load a Java class for instance because of a call to Jess s new function that Object s class loader is used to find it In some Java applications especially in application lE Gamma R Helm R Johnson and J Vlissides Design Patterns Elements of Reusable Object Oriented Soft ware Reading MA Addison Wesley 1995 308 18 1 1 18 1 2 CHAPTER 18 Embedding Jess in Java applications server environments there can be multiple class loaders and using the correct one is important By
549. the bottom Together these nodes form the Rete network and this network is how Jess s working memory is implemented 1 Charles L Forgy Rete A Fast Algorithm for the Many Pattern Many Object Pattern Match Problem Artificial Intelligence 19 1982 17 37 The Rete algorithm 137 At the top of the network the input nodes separate the facts into categories according to their head for example books go through one path and borrowers go through another Inside the network finer discriminations and associations between facts are made until the facts get to the bottom At the bottom of the net work are nodes representing individual rules When a set of facts filters all the way down to the bottom of the network it has passed all the tests on the LHS of a par ticular rule this set together with the rule itself becomes either a new activation record or a command to cancel a previously existing activation record recall that an activation record is an association of a list of facts with a rule that they activate Between the inputs and the outputs the network is composed of two broad categories of nodes one input nodes and two input nodes One input nodes perform tests on individual facts and two input nodes perform tests across multiple facts An example would probably be useful at this point The following rules might be compiled into the network shown in figure 8 1 Jess gt deftemplate x slot a TRUE Jess gt
550. the form for deducting child care expenses won t be needed if the user has no dependents The rules that ask about special circum stances are shown in listing 11 3 Many of these rules are fired unconditionally but a few depend on other answers The rule request childcare expenses won t be activated if the user doesn t have dependents and request reim bursed expenses will fire only if the user has unreimbursed expenses defrule request childcare expenses If the user has dependents answer ident dependents text t amp gt t 0 gt assert ask childcare defrule request employee expenses gt assert ask employee defrule request reimbursed expenses If there were unreimbursed employee expenses answer ident employee text t amp eq t yes gt assert ask reimbursed defrule request moving gt assert ask moving defrule request casualty gt assert ask casualty defrule request on time gt assert ask on time defrule request charity gt assert ask charity defrule request home office gt assert ask home office 11 2 3 11 3 Recommending forms 177 When the interview module executes it asks the user a series of questions and leaves behind a user fact and a collection of answer facts These facts serve as input to the next module recommend which decides which forms the user needs Testing the interview module Te
551. the fuzzy inputs to all the rules in the system executing the rules one at a time and performing a global accumulation of the outputs FuzzyJess does this for you 4 Defuzzify the outputs create crisp numbers from the output fuzzy values 5 Apply the crisp outputs to the system 6 Repeat all the steps until the system is in a controlled state 16 6 2 The Fuzzy HVAC Controller With this brief introduction to fuzzy logic and Fuzzy FuzzyJess we can now return to the HVAC Controller example Recall that the decisions and actions in the rules used to control the heat pumps and vents are all binary set the vent to the open state or crisp if the thermometer reading is less than the desired set point In the two versions of the fuzzy controller described next you will define fuzzy terms for the temperature and for setting the vent positions This indicates that you are going to extend the functionality of the system a little and not restrict the vents to a fully open or fully closed position you will allow them to be partially open as well The code for both versions of the fuzzy HVAC Controller along with all the other code from this book is available from the book s web site Version 1 absolute vent positions In your first attempt at controlling the temperature in the building you will define fuzzy terms for describing the temperature and some terms for controlling the vent opening You ll first modify the Thermometer bean clas
552. things and more The first argument of apply is the name of a function Invoking apply calls that function with all the other expressions as arguments apply lt function name gt lt expression gt The function name can be held in a variable or can itself be the return value of a function In this example the user is prompted for the name of a function which is then called with a list of numbers as arguments the result is then printed The user enters when prompted Jess gt apply read 1 2 3 6 Because the user typed apply added the numbers and displayed the result If the input had been instead the result would have been 4 The apply function therefore turns a bit of data the name of a function into code for Jess to exe cute The eval and build functions take this idea one step further eval and build Because Jess is an interpreted language it doesn t much care where the code it s executing comes from The code can come from a file from the keyboard or 4 4 Defining functions with deffunction 55 even from a variable The eval and build functions convert arbitrary Jess data into Jess code making it simple to write a Jess program that writes other Jess pro grams This means a rule based program can create and incorporate new rules as it runs it can learn This is a powerful feature The eval function accepts a single argument a string containing a complete Jess expression including parentheses
553. tion TransformerFactory tFactory TransformerFactory newlInstance Transformer transformer tFactory newTransformer DOMSource source new DOMSource rules FileWriter writer new FileWriter filename StreamResult result new StreamResult writer try transformer transform source result finally writer close 374 20 6 CHAPTER 20 Jess XML and the enterprise In between reading and writing of course would be code to allow editing of the rules This could be a Swing GUI or it could be a web interface using servlets and JSPs The important thing to remember is that although creating a fully general rule editor might be a difficult undertaking it s much simpler to create a limited editor that allows your users to add rules within a set of constraints appropriate to your application Summary XML is an excellent way to store rules It can easily be translated into the native rule language of almost any rule engine It is easy to parse using one of the many available high quality XML parsers and you can write XML documents easily using the javax xml transform APIs XML can easily be searched using XSLT scripts or from Java code Together these qualities make XML an excellent native format for a rule editor A rule editor that can be used by nonprogrammers is sometimes an important component of an enterprise system In the next chapter we ll examine ways in which rule engines in general an
554. tion 10 6 Finally a recommendation needs a slot to hold a form and perhaps an expla nation deftemplate recommendation slot form slot explanation You ve defined templates named user form question answer and recom mendation It turns out that this collection is sufficient for your needs Let s con sider why these templates are enough Templates you don t need The other possible templates dependent credit and deduction probably won t be a part of the application Looking back at chapter 9 you don t see any thing about the collected knowledge that requires you to store information about individual dependents only the total number of dependents which you ll store in the user template As a result you won t need a dependent template after all The argument for not including credit and deduction is more involved and it s related to an important architectural decision If you stored credits and deduc tions as facts you could write a generic set of rules to operate on these facts The advantage to this architecture is that new forms could be added simply by augment ing the set of credits and deductions that is by asserting new facts You could do this by extending a set of deffacts that would be read at application startup In general adding new facts would be an easier way to add new tax forms than modi fying the rules If you hard coded the credit and deduction information into the rules though then
555. tion creates a simulator and returns it as an RU EXTERNAL_ADDRESS via the Value constructor that accepts a java lang Object argument Note that this class like all the Userfunction classes you will write is not public it is meant to be loaded only via the Userpack age you will define at the end of this section The static getSimulator method is an implementation of a technique men tioned in the previous section It finds an InitSimulator object registered in a given Rete object and extracts and returns its state here the m_simulator mem ber variable Several of the other Userfunctions will use getSimulator to find their associated InitSimulator object and retrieve the Simulator Counting devices As you have seen a single Userfunction class can have multiple personalities It can behave differently depending on the name it was registered with You ll use this trick to write one class to implement both a function to return the number of heat pumps and a second function to return the number of floors in the building package control import jess class CountDevices implements Userfunction static final String N_HEATPUMPS n heatpumps static final String N_FLOORS n floors private String m_name public CountDevices String name m_name name 250 15 5 3 CHAPTER 15 Extending the Jess language public String getName return m_name public Value call ValueVector vv Context c throws JessException Rete en
556. tion of line items Each line item knows which order it belongs to what product is being ordered and in what quantity Although it s not strictly necessary you ll find later that including the customer ID in each line item simpli fies your rules considerably deftemplate order slot customer id slot order number deftemplate line item slot order number slot part number slot customer id slot quantity default 1 Finally recommendations link an order to a recommended product The because multislot lists other products that triggered this recommendation and the type slot distinguishes between product requirements XCON s kind of recommenda tions and marketing recommendations deftemplate recommend slot order number multislot because slot type slot part number With these basic data structures put together you re ready to begin writing the rules for the order configuration system Writing the rules 295 17 5 Writing the rules 17 5 1 The rules you will write in this chapter examine facts and assert recommenda tions but they won t display anything or otherwise interact with a user You ll write the user interfaces for the web based applications in HTML as JSPs and you won t do that until chapter 19 For now to see the effects of the rules you write you ll have to use debugging commands like watch and facts you learned about these in section 6 1 1 As you write these rules
557. to Jess In the next part of this book and in each subsequent part you will develop a nontrivial application Each appli cation is standalone but each part of the book builds on the knowledge and skills developed during the previous parts Part 3 Creating your first rule based application the lax Forms Advisor U a classic expert system is something like being interviewed The pro gram asks you a series of questions and then provides some advice or information In part 3 you ll develop your first nontrivial program to follow this model We ll concentrate on the mechanics of rule based system development There are spe cial sections on knowledge engineering and on testing The system yov ll be developing is called the Tax Forms Advisor Given some information about your personal finances the program will recommend which income tax forms you may need to file It is suitable for installation in a kiosk in the lobby of a post office Collecting the knowledge In this chapter you ll a Learn about knowledge engineering Learn to interview experts a Collect requirements a Assemble domain knowledge 149 150 9 1 9 2 CHAPTER 9 Collecting the knowledge A journey of a thousand miles begins with the first step Lao Tzu The first step in developing any rule based system is collecting the knowledge the system will embody In this chapter your major concern will be to learn how this can be accomplishe
558. to make sure the Rete object is ready and then launches into its own initialization public class Recommend extends BaseServlet public void doGet HttpServletRequest request HttpServletResponse response throws IOException ServletException checkInitialized ServletContext servletContext getServletContext String items String request getParameterValues items String orderNumberString String request getSession getAttribute orderNumber String customerIdString String request getSession getAttribute customerId if items null orderNumberString null customerIdString null dispatch request response index html1 return The call to getParameterValues returns the list of purchased items If the shop ping cart is empty the customer hasn t logged in or there s no order number then Recommend plays it safe and sends the customer to the login screen to start over Note that you always have to use return after you call dispatch otherwise this servlet will try to append its output to the login screen Creating the order Next you need to set up the order in Jess s working memory try Rete engine Rete servletContext getAttribute engine engine executeCommand assert clean up order orderNumberString engine run 19 10 3 The Recommend servlet 345 By asserting the fact clean up order NNNN where NNNN is the order number
559. to the join node s output Finally if a terminal node receives a REMOVE token the corresponding activation record is found and deleted The remaining two tags are more subtle UPDATE is used when a new rule has been added to a preexisting Rete network and the join nodes belonging to that new rule have to be populated with facts The UPDATE tag lets the nodes that already existed know they can safely ignore a token because it s a duplicate of one sent some time in the past this prevents the preexisting nodes from storing duplicate tokens in their memories Finally the CLEAR tag tells the join and ter minal nodes to flush their memories it is used to implement the reset com mand efficiently Easy optimizations for Rete That s it for the basic Rete algorithm There are many optimizations however which can make it even better Two easy ones work by introducing node sharing into the network The first optimization is to share nodes in the pattern network In figure 8 1 there are five nodes across the top although only three are distinct You can modify the network to share these nodes across the two rules The result is shown in figure 8 2 140 CHAPTER 8 Under the hood how Jess works ae LEFT 0 a RIGHT b LEFT 0 a RIGHT b exemple 7 N Figure 8 2 Ea A Rete network that ky orale shares pattern nodes But that s not all the redundancy in the original network Looking at figure 8 2 y
560. tor to create a custom template for a JavaBean 6 5 3 An example JavaBean Now let s work through an example of how a JavaBean can be connected to Jess s working memory Listing 6 2 shows a simple JavaBean class This JavaBean has one property called brightness Note that when you set the brightness property the Bean calls the method adjustTriac that brightens or dims an attached lighting fixture the code for adjustTriac presumably uses the Java Native Interface to access code written in another language like C The value you read using the getBrightness method is thus always tied to the brightness of the light If you had a DimmerSwitch object in Jess s working memory then your rules would be able to reason in real time about the lighting in a building public class DimmerSwitch private int brightness 0 public int getBrightness public void setBrightness brightness b adjustTriac b return brightness int b 90 6 5 4 6 5 5 CHAPTER 6 Representing facts in Jess private void adjustTriac int brightness Code not shown You need a special deftemplate to plug a DimmerSwitch into Jess let s learn how to create one Creating a deftemplate for DimmerSwitch The Jess functions that let you put JavaBeans into working memory are defclass and definstance The defclass function tells Jess to generate a special template to represent a specific JavaBean class whereas definstance puts a shadow fa
561. ts session release The javax rules API is clearly designed to be used in a managed environment like the J2EE For a simple application like the one we re imagining here it could be overkill Consider that by using Jess s native API all the code in this section and the preceding one can be expressed as follows assuming the appropriate defclasses are defined in rules clp Rete session new Rete session executeCommand batch rules clp session definstance java lang Integer new Integer 1 false session definstance java lang String A string false session run Iterator results engine listDefinstances On the other hand both in the J2EE environment and in standalone applica tions javax rules gives you some measure of vendor independence The reference implementation The first implementation of javax rules is the reference implementation RI included as part of the specification This RI is as I mentioned earlier a wrapper for Jess It s an interesting case study because the mapping of the javax rules API forces a specific interpretation onto Jess s more general facilities Here I m not concerned with describing how it is implemented but simply with filling in the missing information that allows you to use it JSR 94 the javax rules API 387 Working memory elements in a RuleSession are arbitrary Java objects Because Jess doesn t let you add arbitrary objects to working memory w
562. ts the unknown name of the person to Fred s right p1 is Fred s unknown position c1 is the unknown color of Fred s pants and p is the unknown golfer s position In these patterns amp means and and means not so name n amp Fred means that this person s name call it n is not Fred Here s the next line Joe is second in line Joe is in position 2 position of Joe is p2 amp 2 amp p1 pants color of Joe is c2 amp c1 Note that you must be careful to read between the lines of the problem as you write this rule You know every golfer is in a different position so you can say with confidence that p2 amp 2 amp p1 Joe s position call it p2 the value of which is 2 is not the same as Bob s position p1 It s possible that p2 and p are the same though Joe might be to Fred s immediate right so you don t mention p here Now the next line of the problem Bob is wearing plaid pants Bob is wearing the plaid pants position of Bob is p3 amp pl amp p amp p2 pants color of Bob amp n is plaid amp c3 amp c1 amp c2 By now you know a lot about Bob s position p3 and pants color 3 You know p3 is not the same as p1 or p2 and you also know it s not the same as p Why Because the golfer in position p wears blue pants and Bob s pants are plaid and the golfers in p1 and p2 are named Fred and Joe not Bob Math class melee 7 Finally you know a lot
563. ty s sake First if a floor is outside the accept able temperature range and its heat pump is off its corresponding heat pump needs to be turned on and put in the correct mode This corresponds to the behavior in truth table 16 2 The rule floor too cold pump off handles the case where a floor is too cold a separate loor too hot pump off rule not shown is also needed Temperature set point defglobal set point 70 Deffunctions deffunction too cold t return lt t set point 2 Rules defrule floor too cold pump off Thermometer floor f reading t amp too cold t HeatPump state off number p amp which pump f gt set hp state p heating 260 16 3 1 CHAPTER 16 Writing the rules There are a few things to notice about this rule We used several of the Userfunc tions defined in the last chapter to keep this rule simple The test number p amp which pump means if this heat pump is the one that services floor You define a deffunction named too cold that deals with the guard line con cept here the lower guard line is defined as 2 degrees below the temperature set point in the defglobal set point Finally you re able to use symbolic names for the state of the heat pump Words like offand heating make the rule eas ier to understand Note that in the test state off you have to enclose the name of the state in quotes so that it s a stri
564. uitable as a text for a university course on practical rule based systems development The course prerequisites should include a course on Java programming The course content would include parts 1 2 and 3 of the book followed by either part 5 or part 6 Additional material could be used as time permits of course Source code downloads The code for all the major examples and applications in this book is available from the book s web site www manning com friedman hill You can also down load a special version of Jess from this web site Typographical conventions This book includes listings of code in both the Jess and Java languages It also con tains transcripts of interactive sessions at the Jess prompt All of these are set in monospace type Keywords function names variable names and symbols in any language are also set in monospace when they occur in the main text It is gener ally clear from context whether I m talking about Jess code or Java code because the two don t look much alike In the interactive session transcripts the Jess prompt and things that you enter are all shown in normal monospace type while responses printed by Jess are shown in italic In step by step examples text that you are to type appears in bold author online Purchase of Jess in Action includes free access to a private web forum run by Man ning Publications where you can make comments about the book ask technical questions and
565. umps using set hp state and open the vents with set vent state You should be able to reduce the temperature of every floor by putting each heat pump in the cooling state Try closing individual vents to see what happens Try to keep each floor at 72 degrees by operating the controls your self it s quite difficult Change the outdoor temperature to 50 degrees by typing into the GUI s text field The temperatures should begin to drop Switch all the heat pumps into the heating state and try to stabilize the temperatures at 68 degrees Not only is this exercise a useful way to test the code you ve written but it may give you some insights into the system s behavior that will be helpful when in the next chapter you write the rules to control it Summary It is easy to extend the Jess language with functions written in Java Any class that implements the jess Userfunction interface can act as a Jess extension func tion Extensions can be loaded from Jess using load function or from Java code using Rete addUserfunction Extensions can be grouped into packages and loaded with a single call to load package Every function built into Jess is itself a Userfunction By implication anything a Jess built in function can do you can do in your own extension functions By way of example you saw how you can manipulate variables and write your own Jess control structures in Java For more useful tools for doing this kind of coding look at the API
566. urse some classes have public member variables like the x and y members of java awt Point which you ll want to both read and modify Instance variables are members of a class that belong to individual objects each object has its own copy of an instance variable Jess can access public instance vari ables of Java objects using the get member and set member functions In this exam ple a Point object is allocated and its x and y members are set and then read Jess gt bind pt new java awt Point lt External Address java awt Point gt Jess gt set member pt x 37 37 Jess gt set member pt y 42 42 Jess gt get member pt x 37 The set member and get member functions also work on class variables There is only a single copy of each class variable and all objects of a class share it Class variables are also called static variables in Java You can access class variables by using the name of the class instead of an object as the first argument to set member Or get member Jess gt get member System out lt External Address java io PrintStream gt Jess gt get member java awt BorderLayout NORTH North Jess converts values for all kinds of member variables between Java and Jess types according to tables 5 1 and 5 2 that is using the same rules as are used for method arguments and return values When you re working with Java objects methods usually return a value to their caller and everything works fine So
567. urthermore Jess can provide the Java main func tion for your program or you can write it yourself You can develop Jess applications with or without GUIs without compiling a single line of Java code You can also write Jess applications that are controlled entirely by Java code you write with a minimum of Jess language code Jess has been deployed in every thing from enterprise applications using J2EE on mainframes to personal pro ductivity applications on handheld devices If you can think of it you can probably implement it with Jess The most important step in developing a Jess application is to choose an architecture from among the almost limitless range of possibilities You must make this choice early in the development of your application One way to orga nize the possibilities is to list them in increasing order of the amount of Java pro gramming involved 38 CHAPTER 3 Introducing Jess 1 Pure Jess language with no Java code 2 Pure Jess language but the program accesses Java APIs 3 Mostly Jess language code but with some custom Java code in the form of new Jess commands written in Java 4 Half Jess language code with a substantial amount of Java code provid ing custom commands and APIs Jess provides the main function the entry point for the program 5 Half Jess language code with a substantial amount of Java code provid ing custom commands and APIs You write the main function 6 Mostly Java code whi
568. usiness depends on Enterprise applications aren t applica tions that individual employees use like Microsoft Word or Adobe Photoshop An enterprise application is run on behalf of the business itself Payroll applications customer databases resource planning applications inventory applications and online ordering applications all fit this definition Enterprise applications have to perform well and be reliable This is tradition ally described in terms of a long list of additional buzzwords Generally enterprise applications are expected to be Scalable As the business grows the application should grow with it You ll typically add more computers for it to run on or upgrade the software infrastructure around it Available Users should be able to get at the application 24 hours a day 7 days a week Downtime should be minimal or nonexistent Transactional It should be possible to undo complete user interactions up until the last moment For example if an ATM machine goes offline while a customer is using it any incomplete transactions should be cancelled cleanly Furthermore many enterprise applications are 20 1 1 20 1 2 Enterprise applications 359 Distributed An enterprise application may consist of individual compo nents running on many different computers sometimes physically sepa rated by great distances Seamless This buzzword usually means that many unrelated components work well together to
569. using only Jess s Java reflec tion capabilities that is you ll write only Jess scripts not Java code In the pro cess we ll look at how Jess lets you develop software using the Java libraries in an iterative and interactive way The GUI you ll develop in this chapter will use Java s Swing library I ll assume some basic knowledge of Swing and Java GUIs in general on your part You might want to keep some documentation for the javax swing package handy for reference Note that you ll be using an iterative approach to develop this GUI you ll see the same Jess functions presented several times with various refinements applied each time The final code for the completed application is available from this book s web site Getting started You ll be using many of the classes in Java s javax swing java awt and java awt event packages Although you could spell out these package names when you need them it will be easier if you use Jess s import function to make this unnecessary Jess s import function like the Java keyword makes it possible for you to refer to all the classes in a given Java package by using their class names without spelling out the package name each time Note how after importing javax swing you can create a javax swing JButton using just its class name Jess gt import javax swing TRUE Jess gt import java awt TRUE Jess gt import java awt event TRUE Jess gt new JB
570. usly track property changes in a Java Bean Jess needs to be notified whenever a property changes in that JavaBean The JavaBean can notify Jess by sending it a special kind of Java event a java beans PropertyChangeEvent Many commercially available JavaBeans already support this kind of notification and you can easily add it to Beans you write yourself If DimmerSwitch offered support for PropertyChangeListeners then it could notify Jess when its brightness changed and the shadow facts could stay in sync with the Beans Let s add that support now Adding PropertyChangeListener support to DimmerSwitch Let s modify the DimmerSwitch class to send PropertyChangeEvents The modi fied class DimmerSwitchDynamic is shown in listing 6 3 The interesting method in DimmerSwitchDynamic is setBrightness This method saves the original value of the brightness member variable then gives it its new value and then sends a property change notification using the PropertyChangeSupport utility class The addPropertyChangeListener and removePropertyChangeListener methods are boilerplate code that is always used with PropertyChangeSupport These methods allow interested external code like Jess to register to be notified when a property changes import java beans public class DimmerSwitchDynamic private int brightness 0 public int getBrightness return brightness Shadow facts 93 public void setBrightness int b int old brightness
571. ution engine Finally once your rule engine decides what rule to fire it has to execute that rule s action part The execution engine is the component of a rule engine that fires the rules In a classical production system such as MYCIN rules could do nothing but add remove and modify facts in the working memory In modern rule engines firing a rule can have a wide range of effects Some modern rule engines like Jess offer a complete programming language you can use to define what happens when a given rule fires The execution engine then represents the environment in which this programming language executes For some systems the execution engine is a language interpreter for others it is a dispatcher that invokes compiled code Developing rule based systems This book is a hands on guide to building useful rule based systems Each indi vidual project in this book covers some aspect of this task presenting realistic examples of every step along the way In this section we look at an overview of the development process we will follow in later chapters Knowledge engineering The first step in the development of any rule based system is to begin collecting the knowledge from which the rules will be derived People who do this for a liv ing are called knowledge engineers Knowledge engineering can be tricky particu larly if the knowledge has to come from human experts Experts aren t always cooperative and even if they are th
572. utral structures instead This would make it possible to radically modify the templates and rules or even to use a dif ferent rule engine without touching the JSPs This encapsulation would have made the sample implementation longer so I didn t use this technique here but it s a good idea from a software engineering standpoint Another related possibility would be to use the javax rules API first men tioned in section 2 5 The javax rules API doesn t offer access to many of the features you used in this application however notably queries are not sup ported We ll look at using javax rules in the last part of this book Finally I have already mentioned the possibility of using a relational database together with Jess to provide long term persistence A database would provide faster more robust access to historical data than the flat file used here Summary In chapter 17 you developed a rule based Recommendations Agent for e com merce The Recommendations Agent rules are the most complex individual rules presented in this book You wrote the rules along with a number of queries that the web application needs to probe Jess s working memory Many of Jess s Java APIs were presented in chapter 18 You saw how Jess is really a programmer s library rather than an application and you learned how to embed Jess in other software In this chapter you embedded the Recommendations Agent in a web applica tion built from three
573. utton Example Button lt External address javax swing JButton gt 13 2 Displaying a window 205 The graphical components of the application will be accessed from different sec tions of Jess code In particular several functions need access to the application s top level window component a JFrame You ll therefore store a reference to it in a defglobal Unfortunately the reset function by default resets the values of defglobals and it would complicate the application if you had to worry about the possibility of your only reference to the JFrame disappearing To deal with this situation you use the set reset globals function to tell Jess to leave defglobals alone when reset is called Jess gt set reset globals FALSE FALSE With these details taken care of you can start coding the GUI Displaying a window The first and most fundamental graphical element you need to create is a javax swing JFrame a top level window Let s create one with an appropriate title size it make it visible and keep a reference to it in a defglobal named frame Jess gt defglobal frame new JFrame Diagnostic Assistant TRUE Jess gt frame setSize 520 140 Jess gt frame setVisible TRUE I m cheating a little here I determined the appropriate size after the whole GUI was designed and then came back here and typed it in In your own programs you will learn the dimensions necessary through experience or e
574. utton OK engine definstance button b true You don t want this defclass to extend another one so you pass nu11 as the last argument to the defclass function Because java awt Button sends Property ChangeEvents you can pass true as the last argument to definstance making this a dynamic instance Calling Jess functions from Java As you ve seen the Rete class exposes many Jess functions as public Java meth ods many but not all Some Jess functions show up as Java methods in other classes For example the Jess gensym function is available as the static gensym function in the jess RU class The Userfunction class that implements gensym is just a thin wrapper around this static Java method Other Jess functions however are implemented directly in the Userfunction class To call them from Java you have to use an instance of the appropriate Userfunction class The helper class jess Funcall can make this process a little simpler As an example there s no specific Java API for the watch function To turn on all the watch diagnostics from Java code the equivalent of watch a11 in Jess you can do this Rete engine Context context engine getGlobalContext Funcall f new Funcall watch engine f arg all f execute context 316 18 5 CHAPTER 18 Embedding Jess in Java applications Once you ve created a Funcall object you can save it and call execute on it any number of times Alternatively
575. vate boolean m_vents m_sensors public Simulator int numberOfFloors if numberOfFloors 3 0 throw new RuntimeException Illegal value m_heatpumps new int numberOfFloors 3 m_vents new boolean numberOfFloors m_sensors new boolean numberOfFloors public int getHeatPumpState int heatPump return m_heatpumps heatPump 1 public void setHeatPumpState int heatPump int state switch state case OFF case HEATING case COOLING m_heatpumps heatPump 1 state break default throw new RuntimeException Illegal value public int getNumberOfHeatPumps return m_heatpumps length public boolean getVentState int vent return m_vents vent 1 public void setVentState int vent boolean state m_vents vent 1 state Defining the hardware interface 227 14 2 3 Simulating getTemperature The meat of the HVAC simulator is the get Temperature method which returns a value based on the complete state of the system Each time you call getTemperature with a given argument you can get a different answer back because the temperatures will constantly change based on many factors At least four things go into the calculation of the temperature on each floor of the building The current temperature on that floor Whether the floor is actively being heated or cooled Heat leakage from outside through the walls Heat leakage from other floors The simulator should t
576. vent to decide what action to take B 7 jess PrettyPrinter This class knows how to format many Jess constructs so that they look nice to a human reader It uses the Visitor pattern so it includes many methods named visit xxx but you won t need to use them To use PrettyPrinter you just need to pass the construct to print as the constructor argument and then call toString to get the formatted version You have to construct a new PrettyPrinter for each construct you need to format public PrettyPrinter Visitable v Argument can be an instance of any of the Jess classes that implement the Visitable interface which include Deffacts Deffunction Defglobal Defquery Defrule and Deftemplate lE Gamma R Helm R Johnson and J Vlissides Design Patterns Elements of Reusable Object Oriented Soft ware Reading MA Addison Wesley 1995 420 APPENDIX B Abridged Java API for Jess public String toString Returns a nicely formatted rendition of the construct that was passed in as a con structor argument B 8 jess Rete B 8 1 This class acts as a facade for all the most important parts of Jess There are meth ods in Rete for managing working memory the agenda the rule base and more In this appendix I ll only mention the most important methods that have no direct equivalent in the Jess language More methods are covered in chapter 18 public Rete Object appObject Constructs a new Rete object The argu
577. vers 2 R Davis B G Buchanan and E H Shortliffe Production Systems as a Representation for a Knowl edge Based Consultation Program Artifical Intelligence 8 1977 15 45 Architecture of a rule based system 19 incorporate a rule engine and most others explicitly or implicitly offer integra tion with one Expert systems really have become ubiquitous we just don t call them by that name anymore 2 3 Architecture of a rule based system The rules in the first expert systems were intertwined with the rest of the soft ware so that developing a new expert system meant starting from the ground up The folks who wrote MYCIN recognizing this fact created a development tool named EMYCIN EMYCIN Empty MYCIN was developed by removing all the medical knowledge from MYCIN leaving behind only a generic framework for rule based systems EMYCIN was the first expert system shell An expert system shell is just the inference engine and other functional parts of an expert system with all the domain specific knowledge removed Most modern rule engines can be seen as more or less specialized expert system shells with features to support operation in specific environments or programming in specific domains This book is about this kind of rule engine A typical rule engine contains m An inference engine Arule base m A working memory The inference engine in turn consists of m A pattern matcher m An agenda m An executio
578. want to replace the server s default error pages with custom ones that hide the true nature of the error but the defaults are handy during debugging The catalog JSP The Catalog servlet packages a product list and forwards it to the catalog JSP for display I m following the convention that Java class names start with an upper case letter whereas JSP filenames start with a lowercase letter hence Cata log java and catalog jsp Because we re not web designers the JSPs are simple and to the point This jsp useBean tag fetches the Iterator that the servlet put into the Http ServletRequest lt HTML gt lt page import jess gt lt jsp useBean id queryResult class java util Iterator scope request gt The next part is pure HTML You have a page title a headline a little text and the header row for the table lt HEAD gt lt TITLE gt Ordering from Tekmart com lt TITLE gt lt HEAD gt lt BODY gt lt H1 gt Tekmart com Catalog lt H1 gt Select the items you wish to purchase and press Check Order to continue lt FORM action Order recommend method POST gt lt TABLE border 1 gt lt TR gt lt TH gt Name lt TH gt lt TH gt Catalog lt TH gt lt TH gt Price lt TH gt lt TH gt Purchase lt TH gt lt TR gt Nothing fancy but it still would be a pain to modify if it were embedded in printi1n statements in a servlet Now comes the real meat of this JSP It uses a Java while l
579. with the necessary rules and facts and then stores it in the ServletContext The BaseServlet class has a few other methods as you ll see One simple method I should mention right now is its implementation of doPost public void doPost HttpServletRequest request HttpServletResponse response throws IOException ServletException doGet request response The doGet method handles HTTP GET requests and doPost handles POST requests This base class method lets you implement one of these and get the other for free Getting the login name Now you ll continue with the implementation of the Catalog servlet s doGet method The name the customer provides on the login screen will be available as a named parameter in the HttpServletRequest argument to the servlet s doGet method in general this is how HTML form data is received by a servlet String customerId String request getParameter customerId if customerId null customerId length 0 19 8 3 The Catalog servlet 339 dispatch request response index html1 return If there s no login name because the customer didn t enter one or because the customer came directly to the Catalog servlet by typing its URL without logging in the servlet invokes the login screen using the utility routine dispatch that you wrote in section 19 5 2 dispatch belongs in BaseServlet because all of your servlets will use it Starting a user session
580. wn Each part presents a rule based system of increasing complexity and also introduces new programming techniques Part 3 presents an information kiosk the Tax Forms Advisor that helps cus tomers choose which income tax forms to bring home You ll learn how to collect expert knowledge and condense it into rules The kiosk as presented has a simple text based interface Part 4 is concerned with the development of the PC Repair Assistant a help desk application with a Swing based graphical interface This example builds on and extends some of the software infrastructure developed for the Tax Forms Advisor In Part 5 IIl guide you through the development of the HVAC Controller an intelligent climate control system for a hypothetical office building This part shows how rule based systems can be interfaced to hardware A special section written by Bob Orchard developer of the FuzzyJ toolkit and the FuzzyJess extension for Jess shows how the HVAC Controller can be enhanced by the use of fuzzy logic Part 6 is about web based e commerce solutions This part presents a Recom mendations Agent that analyzes a customer s past and present purchases to recommend additional items of interest The Recommendations Agent is embedded in a set of servlets and JavaServer Pages in the Tomcat servlet engine Part 7 is a little different The two chapters in this last part cover various top ics relevant to using rule based systems in enterprise appli
581. wnload and automatically install what amounts to a normal desktop application written in Java Java Web Start applications don t have the same kind of lifecycle problems that applets do they can be launched used and exited like a normal application Java Web Start is relatively new but is rapidly gaining acceptance Like applets Java Web Start applications rely on users having a properly configured JVM on their desktop Again it s easier to make this assumption on an intranet than on the Internet 290 17 1 2 CHAPTER 17 Jess on the Web Pros and cons of fat clients Fat client architectures have both advantages and mostly disadvantages The advantages are as follows Little processing power is required on the server because much of the code executes on the client The programmer has maximum control over user interface because it can be written using all of Java s APIs These are the disadvantages It s hard to write complex client software that will work in a wide range of desktop environments Downloading the client software may be intolerably slow for many users especially on the Internet Ifthe client software is installed persistently on the desktop backward compat ibility becomes an issue forcing all users to update software can be difficult The first disadvantage is enough of a problem that fat client architectures are not used very often This problem is most acute for applets because web br
582. xperimentation When you enter the last line of code an empty window appears on the screen as shown in figure 13 1 Diagnostic Assistant lolx Figure 13 1 An empty JFrame created via a short Jess script as shown in the text 206 CHAPTER 13 Adding a graphical interface 13 3 Displaying questions Right now the application you re developing displays questions and accepts answers from the command line As a first step toward producing a real GUI let s change the application to display the text of the questions in the window you ve created for the moment you ll still accept input from the command line You can use a JTextArea inside a JScrol1Pane to accomplish this You need to access the JTextArea from the ask module so you ll store a reference to it ina defglobal named gfield Here s the code to create a JTextArea and add it to your JFrame Jess gt defglobal qfield new JTextArea 5 40 TRUE Jess gt bind scroll new JScrollPane qfield lt External Address javax swing JScrollPane gt Jess gt frame getContentPane add scroll Jess gt qfield setText Please wait Jess gt frame repaint You set the text of the new text area to the string Please wait Figure 13 2 shows what the window looks like now As you ll recall from chapter 12 you used the ask module as originally defined in chapter 10 to provide the temporary command line interface to your application In th
583. y rules Prints a message when any rule fires compilations Prints a message when any rule is compiled activations Prints a message when any rule is activated or deactivated showing which facts have caused the event facts Prints a message whenever a fact is asserted or retracted focus Prints a message for each change to the module focus stack a all All of the above 414 APPENDIX A Jess functions while lt boolean gt do lt expression gt Allows conditional looping Evaluates the Boolean expression repeatedly As long as it does not equal FALSE the list of other expressions is evaluated The value of the last expression evaluated is the return value The symbol do after the Boolean expression is optional If present it is ignored Abridged Java API for fess 416 B 1 B 2 APPENDIX B Abridged Java API for Jess The Jess library contains about 75 public classes Many of them are used only rarely by Jess programmers Even among the most used classes plenty of methods are used very infrequently This appendix lists only the most important methods of the most important classes a 20 000 foot view that should be a big help to any one just getting started with the Jess library jess Context This class represents the execution context of a function If you write your own Jess functions in Java you get a Context object as an argument to the call method The Context gives you access to the Re
584. y about them when you need them Create the directory structure Each web application should have its own subdirectory in TOMCAT webapps This application directory should in turn have a subdirectory WEB INF and WEB INF should have a classes subdirectory You might create more subdirectories but this small set is enough for now Copy the files The web xml file goes into the WEB INF directory The Hello class file goes into classes The resulting structure looks like this TOMCAT webapps Hello WEB INF web xml classes Hello class 328 CHAPTER 19 Deploying web based applications It s not complicated but you do have to get it right During development you must do this many many times so you should write a small script or batch file or Ant task or Makefile target to do this automatically My Makefile for this servlet looks like this SUFFIXES java class JAVA_HOME usr java j2sdk1 4 0 TOMCAT usr local jakarta tomcat APPNAME Hello SRC shell find name java print OBJ SRC java class build OBJ jJava class javac classpath TOMCAT common lib servlet jar lt install build rm rf TOMCAT webapps APPNAME mkdir p TOMCAT webapps APPNAME WEB INF cp web xml TOMCAT webapps APPNAME WEB INF mkdir p TOMCAT webapps APPNAME WEB INF classes cp class TOMCAT webapps APPNAME WEB INF classes clea
585. y construct can use virtually all the same features that defrule LHSs can including all the special conditional elements described in this chapter The function ppdefrule can also pretty print queries Jess treats a defquery as a special kind of defrule in many contexts for instance the rules command lists defquerys as well as defrules As you can see the run query function lets you pass parameters to a query you passed numbers representing the upper and lower limits of a price range to the find affordable gifts query Lets examine this process a little more closely The variable declaration You might have already realized that two different kinds of variables can appear in a query those that are internal to the query like p in find affordable gifts and those that are external or to be specified in the run query command when the query is executed Jess assumes all variables in a query are internal by default you must declare any external variables explicitly using this syntax declare variables x y When you invoke a query using the run query function you must supply exactly as many variables as are listed in the variables declaration Some queries may not have any external variables in this case the variables declaration is optional Query trigger facts When Jess compiles a defquery it inserts an extra pattern as the first one in the query This first pattern is of the form __ query trigger name x y where
586. y rule based system is concerned with some subset of all the world s collected knowledge This subset is called the domain of the system The process of collecting information about a domain for use in a rule based system is called knowledge engi neering and people who do this for a living are called knowledge engineers On small projects the programmers themselves might do all the knowledge engineering whereas very large projects might include a team of dedicated knowledge engineers 9 2 1 Introduction to knowledge engineering 151 Professional knowledge engineers may have degrees in a range of disciplines obvious ones like computer science or psychology and domain related ones like physics chemistry or mathematics Obviously it helps if the knowledge engineer knows a lot about rule based systems although she doesn t have to be a programmer A good knowledge engineer has to be a jack of all trades because knowledge engineering is really just learning the knowledge engineer must learn a lot about the domain in which the proposed system will operate A knowledge engi neer doesn t need to become an expert although that sometimes happens But the knowledge engineer does have to learn something about the topic In gen eral this information will include The requirements Looking at the problem the system needs to solve is the first step However you might not fully understand the problem until later in the process The pr
587. y the mileage slot of the used car template runtime errors would occur whenever candidate was bound to a new car fact because the new car template doesn t have such a slot If the RHS of a rule uses a variable defined by matching on the LHS of that rule and the variable is defined by one or more branches of an or pattern but not all branches then a runtime error may occur For example if the RHS of 110 CHAPTER 7 Writing rules in Jess might buy car used the variable m which is defined only when the rule matches a used car fact then when it fired in response to a new car fact you d see an error message and Jess would stop firing rules The and group can be used inside an or group and vice versa In the latter case Jess rearranges the patterns so that there is a single or at the top level For example the rule Jess gt defrule prepare sandwich and or mustard mayo bread gt is automatically rearranged as follows Jess gt defrule prepare sandwich or and mustard bread and mayo bread gt Jess rearranges the patterns of any rule that has or conditional elements in it so that in the end there is at most one or per rule and it appears at the top level Jess may use DeMorgan s rules to accomplish this result DeMorgan s rules are a set of two formulas that describe legal ways of substituting logical expressions Written in Jess syntax they can be stated as follows not or x y
588. y the status code to be passed to System exit Youll fix that in the next section Handling arguments The first parameter to call is a jess ValueVector The jess ValueVector class is Jess s internal representation of a list and therefore it has a central role in pro gramming with Jess in Java The ValueVector class is used to represent plain lists whereas specialized subclasses are used as function calls jess Funcall and facts jess Fact Other classes like jess Deftemplate use ValueVectors internally Working with ValueVector is simple Its API is reminiscent of standard con tainer classes like java util Vector Like that class it is a self extending array When new elements are added the ValueVector grows in size to accommodate them Here is a bit of example Java code in which you create the Jess list abc ValueVector vv new ValueVector vv add new Value a RU ATOM vv add new Value b RU ATOM vv add new Value c RU ATOM The get method retrieves an element from the ValueVector using a zero based index Continuing the previous example Prints b System out println vv get 1 When your Userfunction s call method is invoked the first element of the val ueVector argument is the name of your function and the subsequent elements are the arguments that were passed to your function How many arguments You can get the total length of the ValueVector using the size method It fol lows that
589. ynamic For example to create the times tamp you could use default dynamic time The slots we ve looked at so far have all contained single values Slots that hold multiple values are useful too You ll learn how to create them next Multislots The normal slots we ve looked at so far can each hold only a single value Some times though it s handy to keep a list of things in a slot For example if you wanted to keep track of a person s hobbies in a hobbies slot you d need to be able to handle people who have more than one way to spend their free time You can create slots that can hold multiple values by using the multislot keyword Jess gt clear TRUE Jess gt deftemplate person People in actuarial database slot name default OCCUPANT slot age slot gender multislot hobbies TRUE Jess gt assert person name Jane Doe age 22 hobbies snowboarding restoring antiques gender Female lt Fact 0 gt Now Jane has two hobbies both listed in the single hobbies multislot If you don t specify a default value for a multislot Jess uses an empty list Usu ally this is what you want but you can specify a different default if you need to The values contained in the slots of an unordered fact are not fixed you can change them whenever you want using the modify function 6 3 4 6 3 5 Unordered facts 85 Changing slot values with modify Very often a rule acts on a fact by updatin
590. you re drawing flowcharts during an interview with a domain expert software might be too unwieldy in which case a dime store plastic drawing tem plate and a sharp pencil might be a better choice From flowcharts to rules Translating flowcharts into rules is fairly straightforward In general there is one rule for every action box and there are additional rules for some of the oval boxes A possible set of rules for the flowchart in figure 12 1 is shown in listing 12 1 The listing includes three rules one each for the action boxes in the flowchart and one for the END oval Note that each rule corresponds to a specific hypothesis about what might be wrong with the car In a more complex flowchart this last correspondence might no longer hold but the matchup between rules and boxes still would START j 1 Car Fene no has tuei gt C Buy fuel YES YES Li Take car Figure 12 1 to mechanic A simple flowchart representing a trivial diagnostic process The hypothesis The car won t run t s because it has no fuel is implicit in EnD ST the diagram defrule no fuel This rule corresponds to the buy fuel box car starts no 190 12 2 CHAPTER 12 Writing the PC Repair Assistant car has fuel no gt assert buy fuel defrule faulty engine This rule is for the take car to mechanic box car starts no car has fuel yes assert take car to mechanic defrule car
591. you move the temperature set point and redefine the tem perature FuzzySet Each time through the loop you update the variable that represents the rate of temperature change You also need to change the way in which you set the vent position Instead of try ing to set an absolute value for the valve position you ll determine how much to change the valve position from its current position To do so you define the seven terms NB NM NS Z PS PM and PB representing a Negative Big change a Negative Medium change a Negative Small change a Zero change a Positive Small change and so on The final changes to the Vent class are shown in listing 16 4 the new fuzzy variable represents the magnitude of the desired change to the vent position package fuzzy import nrc fuzzy public class Vent extends BeanSupport private Hardware m_hardware private int m_floor private static FuzzyVariable m_ventChangeFV null static try m_ventChangeFV new FuzzyVariable ventChange 1 1 catch InvalidFuzzyVariableNameException nameEx 280 CHAPTER 16 Writing the rules NOTHING catch InvalidUODRangeException rangeEx NOTHING try m_ventChangeFV addTerm NB m_ventChangeFV addTerm NM m_ventChangeFV addTerm NS m_ventChangeFV addTerm Z m_ventChangeFV addTerm PS m_ventChangeFV addTerm PM m_ventChangeFV addTerm PB new new new new new new new catch XValuesOutOfO
592. you write rules to match the trigger facts The ask mod ule asks questions based on the MAIN ask facts it sees so your backward chain ing rule called supply answers can match MAIN need answer trigger facts and assert MAIN ask facts in response this process is shown in figure 12 3 The rule supply answers is shown in listing 12 3 supply answers has the auto focus property and is in its own module this is the same trick the ask module uses so that it can fire immediately whenever it is activated 194 12 4 CHAPTER 12 Writing the PC Repair Assistant defrule power supply broken answer ident sound text none b need answer ident sound defrule ask question by id ask id assert answer ident id text reply defrule supply answers need answer ident id assert ask id Figure 12 3 The cycle of rules that fire to get answers from the user in this application The rule power supply broken needs an answer fact before it matches Jess s backward chaining engine detects this requirement and asserts a need answer fact The rule supply answers detects this fact and asserts an ask fact in response Finally the ask question by id rule fires in response to the ask fact and supplies the answer fact needed by power supply broken do backward chaining answer defmodule trigger defrule trigger supply answers declare auto focus TRUE MAIN need answer ident id not MAIN answer id

Jess in Action e-book

Contents

Download Pdf Manuals

Related Search

Related Contents