The RQL v1.5 User Manual
Contents
1. The IST Programme Key Action 3 Action Line IST 2000 3 1 4 Contract number IST 2000 26074 Deliverable D3 2 1 The RQL v1 5 User Manual Version Date Authors Contributors Distribution v1 5 14 Feb 2002 Grigoris Karvounarakis Vassilis Christophides ICS FORTH Dimitris Plexousakis Sofia Alexaki ICS FORTH PUBLIC The RQL v1 5 User Manual Grigoris Karvounarakis Vassilis Christophides Institute of Computer Science FORTH Vassilika Vouton P O Box 1385 GR 711 10 Heraklion Greece gregkar christop Gics forth gr February 14 2002 Abstract This document is a tutorial for the RDF Query Language RQL RQL is a typed language following a functional approach that is defined by a set of basic queries and iterators This tutorial presents examples of basic meta schema schema and data queries as well as select from where filters iterators containing generalized path expressions and shows how such queries can be nested to form more complex queries 1 Introduction This document is a tutorial for the RDF Query Language RQL RQL is a typed language following a functional approach a la OQL 2 It is defined by a set of basic queries and iterators These basic queries are the building blocks of the query language and will be the first to be presented Then such queries and iterators can be used to build more complex queries through functional composition by preserving type integrity constraints which2. as in the following query Q2 3 5 SELECT P Y FROM Painter P Y The result of this query is depicted in Figure 4 Similar to the use of class names in path extremeties as in the above queries we can also use literal type names where applicable e g to find all string valued properties Q2 3 6 SELECT P FROM OP string 2 4 Schema Navigation Simple path expressions as the ones presented in previous queries can be composed using 9 the operator in order to traverse paths in schema graphs Consider for example the 10 following query Q2 4 1 Find the ranges of the property exhibited that can be reached from a class in the range of property creates SELECT X Z FROM creates X Y exhibited Z In this query X Y and Z iterate over the subclass tree of the range or domain and range of the properties creates and exhibited respectively Then the composition of the two is a shortcut notation implying the condition X Y Thus an equivalent paths using to the above query without is the following Q2 4 1 1 SELECT X Z FROM creates X Y exhibited Z WHERE X Y Since we do not want to retrieve classes at the domain of exhibited we can omit the Y from the above query Q2 4 2 SELECT X Z FROM creates X exhibited Z The implied condition for this query is slightly different instead of computing the subclass tree of the domain of exhi
3. Z see fnmatch man page at http www europe redhat com documentation man pages man3 fnmatch 3 php3 15 resource atring httpif Z www rodin fr Crucifixian Figure 7 The result of query Q3 3 3 As should become more clear with this query the implied join condition does not involve at all any schema relationship between the two paths Indeed in this example last modified is defined on class ExtResource and not on Museum Still since there are several resources that are multiply classified under both classes and several of them have a modification date defined this query returns meaningful results In Section 4 we are going to illustrate how one can also filter resource desciptions according to schema restrictions Note that when using constant class or property names as the path s elements e g Museum last modified path components are automatically considered as data paths if no variables are defined on their extremeties On the other hand paths containing property P or class X variables as their elements are treated as schema paths i e their domain or range is used to infer the implied condition As a result the following is a pure data path Q3 3 2 Find the titles of the resources where are exhibited the resources that have been created by a Sculptor SELECT Z FROM Sculptor creates exhibited title Z This is equivalent with the following Q3 3 2 1 SELECT Z FROM Sculptor A B creates C D
4. are specific for each operation allowing arbitrary nesting in a query RQL supports generalized path expressions 3 4 1 featuring variables on labels for both nodes i e classes and edges ie properties which will also be illustrated through examples and provide the basis for more complex queries cThesaus gt Enumera fits Class DS vc WebResourte gt ww ocle arg schema rdf w Icom coms exhibited fame Artifact Museum DA wn r wa Te Nigeation s upts is ee location Thesaurts last_modified Dale time q T Sculpture i s lt A mr CEResource a i paints lL technique s Een ttle Painter gt Pairting A K a ig Cubist Flomish l y mae fechnique P Iname 7 y i J titte E n i tire 7 mn 2 Michelangela tame sculpts ik a _ exhikiteg MARE fi it Suonarroti c f q technique G i 5 technique Q ki exhibited i amp amp r1 itle amp Iname Rodin Museum mame r1 wew cuture netipicasso 132 cuca CSaet eee r9 uww culture net Rembrand 710 http 4www artchive com rembrendtertist_at_his_eazel jpg 111 http 4www artchive com rembrendt 2braham jpg ri www culture nevrodind2d 713 http 4www artchive combrucifixion ipg ri4http 4www rodin fr 15 tpt cg 1 com nmusees ecomuses _ S htipuilwew muzeum esdvornan gi O htip twew museu
5. consider an extended class or property in terpretation that is the union of the set of proper instances of a class with those of all its subclasses In order to obtain only the proper instances of a class i e only the nodes labeled with the class URI RQL provides the special operator Then to find only proper instances of class Artist we can issue the following query 12 Q3 1 2 Artist In our example the result is the empty bag since no resource has been directly classified as instance of Artist Additionally RQL uses as entry points to an RDF description base not only the names of classes but also the names of properties as for example creates By considering properties as binary relations this basic query will return the bag of ordered pairs of resources belonging to the extended interpretation of creates Q3 1 3 creates An equivalent RQL query employing a select from where filter is the following Q3 1 3 1 SELECT X Y FROM X creates Y Similarly with the example of classes we can omit pairs of resources that are connected through subproperties of creates i e find only proper instances of creates Q3 1 4 creates 3 2 Filtering Resource Descriptions For data filtering RQL relies on standard Boolean predicates as lt gt and like for string pattern matching All operators can be applied on literal values i e strings integers reals dates or resource URIs and class prop
6. definitions in the former respre sent RealWorldObjects while the latter instantiates the metaclass WebResource The lower part depicts superimposed resource descriptions created for several Museum Web sites and artifacts available on the Web according to these schemas 2 Schema Querying 2 1 Basic Schema Queries In order to traverse class property hierarchies defined in a schema RQL provides functions such as subClassOf for transitive subclasses and subClassOf for direct subclasses For example we can issue the queries Q2 1 1 subClassOf Artist subClassOf Artist to find all transitive direct subclasses of class Artist Similarly functions superClassOf superClassOf return transitive direct superclasses Similar functions exist for schema properties i e subProperty0f and subPropertyDf For example we can ask for all transitive direct subproperties of creates Q2 1 2 subPropertyOf creates subProperty0f creates Similarly functions superProperty0f superProperty0f return superproperties All these functions may also be used with a second integer parameter in order to return e g subclasses of Artist up to depth 2 Q2 1 3 subClassOf Artist 2 Then for a specific property we can find its definition by applying the functions domain and range Q2 1 4 domain creates range creates Moreover RQL provides the function namespace in order to retrieve the namespace prefix i e
7. 01 01 14 resource date httpif www museur es 2000 06 09715 30 34 00 00 resource date http fwww radin fr 2000 01 31716 59 00 00 00 Figure 6 The result of query Q3 3 1 Q3 2 7 Find resources whose working_hours are at least one of 9 1 5 8 9 4 9 9 values of enumerated types are treated as strings SELECT X Y FROM X working_hours Y WHERE Y in bag 9 1 5 8 9 4 9 9 Q3 2 8 Find the resources whose location is equal or a narrower term than FRANCE in the localization thesaurus range of property location SELECT X Y FROM X location Y WHERE Y lt FRANCE 3 3 Navigating in Description Graphs In a way similar to schema navigation as presented in Section 2 4 we can compose data path expressions in order to navigate in description graphs Consider e g the following query Q3 3 1 Find the Museum resources and their modification date SELECT X Y FROM Museum X Z last_modified Y The result of this query is shown in Figure 6 Note that in this path expression we compose a path component with a class element Museum with another with a property element last modified Composition of data path expressions is equivalent with a join between the collections of the corresponding path elements i e the extent of Museum and last modified This join is made explicitly in the following query equivalent to Q3 3 1 Q3 3 1 1 SELECT X Y FROM Museum X Z last_modified Y WHERE X
8. Flemish creates Y Painting In this query we are using node labels i e class names in order to restrict the source and target values of a property More precisely the implied conditions in this path expression are i Flemish should be a subclass of the domain of creates ii Painting should be a subclass of the range of creates iii X should belong to the extent of Flemish and iv Y should belong in the extent of Painting The result of this query is illustrated in Figure 8 The difference between mixed paths and e g data paths should be more obvious in the next example Consider that we want to find all the Painting resources that have been exhibited as well as the related target resources of type ExtResource Note that the class ExtResource does not appear in the range of property exhibited The appropriate query is the following 17 resource are ulture net f michelangela http www photojournal com classicart y ital masters michelangela faculpturfdescent jpg resource resource www culture net mich lang ela http www photajournal cam classicart ital masters michelangela faculptur theslave jpg resource resource www culture net fradind2d http fwww artchive caom crucifizian jpg Figure 9 The result of query Q5 2 Q4 2 SELECT X Y FROM 4 X Painting texhibited Y ExtResource Z The join condition between the two paths is obviously inferred as that of a data path composition ie Y Z This means tha
9. Proc of ACM SIGMOD pages 413 422 1996 F van Harmelen P Patel Schneider and I Horrocks Reference description of the DAML OIL ontology markup language http www daml org 2001 03 reference html March 2001 22
10. bited we merely require that X is a subclass of the domain of exhibited Q2 4 2 1 SELECT X Z FROM creates X exhibited Z WHERE X lt domain exhibited A similar path expression returning all properties that can be applied on range classes of property creates is the following Q2 4 3 SELECT X P range P FROM creates X P Figure 5 summarizes different combinations of the composition of two schema path com ponents having properties or property variables as their elements 11 Path Expression Variable bindings Pi X P2 X T Pi X P3 X T3 Pi C3 X P2 X T3 Pi X C3 P2 X T3 Pi P P P P P X P X P T1 P2 T3 P3 P1 C3 P P P P Q P Q P1 P2 P1 P3 T3 hla Figure 5 Summary of RQL Schema Path Expressions 3 Querying Resource Descriptions The core RQL data queries essentially provide the means to access RDF description bases with minimal knowledge of the employed schema s 3 1 Basic Data Queries We can access any RDF collection class or property extents container values with RDF data or schema information etc by just writing its name This is the case of RDF classes considered as unary relations Q3 1 1 Artist An equivalent RQL query employing a select from where filter is the following Q3 1 1 1 SELECT X FROM Artist X It should be stressed that by default we
11. e class Artist has been defined SELECT namespace C FROM Class C WHERE C like Artist Q8 2 Find the description of resources excluding desciptions related to classes and properties of namespace ns1 SELECT X SELECT C SELECT P Y FROM W C P Y WHERE namespace C nsi and namespace P ns1 FROM C X FROM Resource X USING NAMESPACE nsi amp http 139 91 183 30 9090 RDF VRP Examples demo admin rdf 9 Nested Queries As we mentioned in the beginning of this tutorial RQL being a functional language allows arbitrary composition of simple queries and iterators in order to form more complex queries For example we can nest any RQL functions Q9 1 Find the subclasses of the range of the property creates subclass range creates Moreover RQL allows the introduction of nested queries in any of the select from where clauses The following examples depict this functionality Q9 2 Find the description of the resource with URI http www museum es i e group properties by the class under which the resource is classified nesting in the SELECT clause SELECT C SELECT P Y FROM Z D P Y WHERE Z X and D C FROM C X WHERE X amp http www museum es 20 claas integer Sculptor 4 claag integer Painter 4 claaa integer Flemish 4 class integer Cubist 4 Figure 10 The result of query Q9 5 Q9 3 Find the subclasses of Artist except subclasses of Sculptor as
12. erty metaclass names For example Q3 2 1 are conditions between integers It should be stressed that this also covers comparisons between class or property names e g Q3 2 2 Painter lt Artist 13 This is equivalent to the basic boolean query Q3 2 2 1 Painter in subclassof Artist Disambiguation is performed in each case by examining the type of operands e g literal value vs URI equality lexicographical vs class ordering etc Finally the second parameter of the operator like is always a wildcard expressions i e a character string that may also contain the special characters etc The first parameter may be either a string thesaurus term or enumeration value or a schema name class property etc For example Q3 2 3 foobar like foox Artist like Art The following are a few example queries illustrating filtering conditions on various data types Q3 2 4 Find the file_size of the resource with URL http www artchive com rembrandt abraham jpg SELECT Y FROM 4 X file_size Y WHERE X amp http www artchive com rembrandt abraham jpg Q3 2 5 Find the titles of resources whose URL matches www artchive com i e contains the sub string www artchive com SELECT X FROM 4 X title Y WHERE X like www artchive com Q3 2 6 Find resources that where modified after year 2000 and their modification date SELECT X Y FROM 4 X last_modified Y WHERE Y gt 2000
13. exhibited E F title Z WHERE A B and C D andE F However this shortcut notation for path compositions does not accomodate tree like path expressions In this case composition conditions have to be expressed explicitly as in the following query Q3 3 3 Find the titles of exhibited resources that have been created by a Sculptor as well as the resources where they are exhibited SELECT Z W FROM Sculptor creates Y exhibited Z V title W WHERE Y V The result of this query is shown in Figure 7 16 ne ue T culture net f rerb randt pattie www arte hiwe cir rerb randt artiat_ at hia emel gag Pe Woe ne uc www c Ult ur et f rerb randt Att A ae arte ied cif rer randt ab reba Figure 8 The result of query Q4 1 For such cases one can use the facility of re using already defined variables as described in Q2 3 2 2 to imply again a join condition Q3 3 3 1 SELECT Z W FROM Sculptor creates Y exhibited Z Y title w 4 Using Schema to Filter Resource Descriptions Up to now we have seen how we can query and navigate in schemas as well as how we can query and navigate in description graphs regardless of the underlying schema s Still RQL allows to combine schema and data filtering and navigation through the use of mixed path expressions Consider for example the following query in contrast to Q2 3 2 and Q3 1 3 1 Q4 1 Find the Flemish resources that have created Painting resources SELECT X Y FROM 4 X
14. ied on Metaclasses e g Q2 2 3 subClassOf Class subClassOf Property while domain range may return metaclasses that may contain classes properties thesauri etc for properties defined between such metaclasses Q2 2 4 domain related range related In order to be able to retrieve only data properties that are defined at schema level illustrated at the middle part of Figure 1 i e relations between resources or attributes of resources one can use the special built in metaclass DProperty Q2 2 5 DProperty Finally user defined metaclasses see upper part of Figure 1 can also be used as basic queries to retrieve the schema classes or properties which are their instances e g RealWorldObject SchemaProperty 2 3 Class amp Property Querying RQL also provides a select from where filter in order to iterate over these collections by introducing variables Given that the whole description base or related schemas can be viewed as a collection of nodes edges path expressions can be used in RQL filters for the traversal of RDF graphs at arbitrary depths A first use case of such queries is schema browsing or filtering this is especially useful for real scale applications which employ large description schemas For example we can get both the direct domain and range of a schema property using the query Q2 3 1 seq domain creates range creates In this query we use the sequence constructor o
15. m esquemica jpg r4 htip dwow museum es uP rope rtyOrli sA JS wew cuture netfnicheangelo 6 htip dwew photojcumal comblaszicata almasterdnichelangelo sculptur descent jpg r7 chtip wew photojcumal comblassicart aalmastersnichelangelo sculptantheslave jpg Figure 1 An example of RDF schemas and resource descriptions for a Cultural Portal In this manual we are going to illustrate the use of RQL through queries of increasing complexity also separating them according to different use cases that they satisfy As a running example for the RQL queries we will use a Cultural Portal containing descriptions about resources such as Museum Web sites or Web pages with exhibited artifacts both from a Portal administrator and a museum specialist perspective i e described according to the corresponding RDF schemas Figure 1 depicts both the employed schemas upper part and the RDF descriptions lower part contained in this example description base The upper part of this figure consists of the default RDFS meta schema classes i e Class and Property as well as user defined metaclasses specializing them i e Real WorldOb ject WebResource and SchemaProperty Moreover it contains two meta schema properties namely related whihc connects classes and maxCardinality which is defined on properties and has an integer value The middle part consists of two schemas intended for museum specialists and Portal administrators respectively Class
16. ng property class paints Painting property class aculpta Sculpture Figure 3 The result of query Q2 3 4 because they are inherited from a superclass of Painter Thus an equivalent RQL query is the following Q2 3 3 1 SELECT P range P FROM DProperty P WHERE domain P gt Painter In this query the gt predicate in the WHERE clause denotes class ordering using the sub classof relation as we will explain later In order to retrieve only properties that are directly defined on Painter we simply have to replace the gt with in the WHERE clause of Q2 3 3 1 Suppose now that we only want to find relationships that can be applied on Painter i e only properties with class range and iterate on their subclasses i e get all possible range classes For this we can use the following query Q2 3 4 SELECT P Y FROM 4 Painter P Y The use of a schema variable with the prefix restricts the result to the properties with class range More specifically it implies the condition range P in Class The result of this query is shown in figure 3 property literal firat name string property literal l t name string property class Artifact property clas Sculpture claas Painting property clas paints Painting Figure 4 The result of query Q2 3 5 In order to also include properties with literal thesauri enumeration or metaclass range we use the prefix
17. perator seq to construct a sequence value explicitly Then consider for instance the following query where given a specific schema property we want to find all related schema classes Q2 3 2 Which classes can appear as domain and range of the property creates SELECT C1 C2 FROM Ci creates C2 In this query we use the prefix for variable names to denote variables ranging on schema classes i e node labels The result of this query is shown in Figure 2 A similar query not including the direct domain and range of property creates is the following Q2 3 2 1 SELECT X Y FROM subclassof domain creates X subclassof range creates Y A more explicit but also less efficient query that is equivalent to Q3 2 is the following Q2 3 2 2 SELECT X Y FROM Class X Class Y X creates Y Since X denotes class variables it is equivalent with a scan on Classes ClassX Then we can re use already defined variables X Y instead of defining new schema variables in the extremeties of creates This way we implicitly apply a condition on X and Y X Y should be class class Artist Artifact class class Artist Sculpture class class Artist Painting class class Sculptor Artifact class class Painter Painting class class Flemish Artifact class class Flemish Sculpture class class Flemish Painting class class Cubist Artifact class class Sculptor Sculpture clas
18. s class Cubist Sculpture class class Cubist Painting class class Sculptor Painting class class Painter Artifact class class Painter Sculpture Figure 2 The result of query Q2 3 2 a subclass of the domain range of property creates Moreover in order to disambiguate such cases where we don t use schema variables but we refer to schema operations on the domain and range of properties from data paths where we want from and to values of a property we use the symbol as a position indicator This will be more obvious in examples of mixed paths in Section 4 where is used to separate data and schema variables when they both appear at the same end of the path Similarly a constant class name can be used instead of an already bound variable in order to find only properties that can applied on class Painter Q2 3 3 Find all properties defined on class Painter and its superclasses i e all properties that can be applied on class Painter SELECT P range P FROM Painter P In this query QP denotes a property variable that is implicitly range restricted on the set of all data properties i e DProperty Then the result of Q2 3 3 contains all properties that may be applied on Painters either because they are directly defined on class Painter or property class echibited Museu rri property clea creates Artifact property clas creates Sculpture property class creates Painti
19. s not exist the query will return a runtime error Alternatively the Boolean operator in can be used for membership test in Bags For example Q6 1 seq domain creates range creates 0 returns the first element of the sequence while Q6 2 tt 1 in bag 1 2 tt amp www culture net picasso132 in Painter are true since the first bag contains 1 while the resource www culture net picasso132 belongs to the extent of Painting 7 Aggregate Functions Last but not least RQL is equipped with a complete set of aggregate functions min max avg sum and count For instance we can inspect the cardinality of class extents or of bags using the count function Q7 1 count Painting Note that the parameter of aggregate functions may be any query returning a collection of a proper type as in the following query 7 2 Find the maximum number of direct subclasses max SELECT count subclassof C FROM Class C WHERE C Resource 19 8 Namespace Queries For cases when several different schemas are used at the same time RQL provides the op eration namespace returning the namespace prefix of its operand as well as an extra USING NAMESPACE clause allowing the definition of namespace prefix variables which can then be used inside queries in order to disambiguate cases when the same e g class name appears in several schemas The following queries depict this kind of functionality Q8 1 Find all namespaces wher
20. t using the above expression we do not enforce the range of exhibited to be related with the class ExtResource If we wanted to pose such a restriction the query would be similar to Q4 1 Q4 2 1 SELECT X Y FROM 4 X Painting exhibited Y ExtResource Since ExtResource is not a subclass of the range of property exhibited this query would produce a warning and return an empty result 5 Set based Queries Common set operators union intersect minus applied to collections of the same type are also supported For example the query Q5 1 Sculpture intersect ExtResource will return a bag with the URIs of all resources which are classified under both Sculpture and ExtResource Moreover we can mix collections denoted by schema names with others created by queries Q5 2 creates minus SELECT X Y FROM X paints Y The result of this query is illustrated in Figure 9 18 However the following query will return a type error since the function range is defined on names of properties and not on names of classes Q5 3 bag range Artist union subclassof Artifact 6 Other Container Queries RQL also allows the manipulation of RDF container values More precisely we can explicitly construct Bags and Sequences using the basic RQL queries bag and seq as we did in Q3 1 To access a member of a Sequence we can use as usual the operator with an appropriate position index If the specified member element doe
21. the URL of the schema where it is defined of any schema name i e class property metaclass etc Note that this function returns only one namespace and works only if it s parameter corresponds to a uniquely identified name For example the query Q2 1 5 namespace Artist works only if there is only one class with name Artist in all schemas that have been loaded For cases when same names are used in different schemas one can use the using namespace clause in order to resolve such naming conflicts explicitly e g Q2 1 6 namespace ns Artist USING NAMESPACE ns amp http 139 91 183 30 9090 RDF VRP Examples demo culture rdf In Section 8 we are also going to depict how one can find all namespaces in which e g Artist has been defined 2 2 Querying the Meta schema More generally the whole schema can be queried as normal data using appropriately defined meta classes This is the case of the default RDF classes Class and Property Using these names as basic RQL queries we will obtain in our example the names of all the classes and properties illustrated in the middle part of Figure 1 Q2 2 1 Class Property Moreover we can use other meta schema names as collection names to get their contents e g Q2 2 2 Literal Thesaurus Enumeration In order to also accomodate user defined metaschemas e g DAML OIL 5 we have overloaded functions such as subClassOf subClassOf so that they can also be appl
22. well as their corre sponding direct instances nesting in the FROM clause SELECT Y SELECT Z FROM Y Z FROM SELECT X FROM Artist X minus SELECT X FROM Sculptor X Y Q9 4 Find the most recently modified resource nesting in the WHERE clause SELECT X Y FROM X last_modified Y WHERE Y max SELECT Z FROM Whlast_modified Z Finally the following query involves several nested queries and aggregate functions Q9 5 Find the classes with the maximun number of properties SELECT C count SELECT P FROM C P FROM Class C WHERE C Resource and count SELECT P FROM C P max SELECT count SELECT Q FROM D Q FROM Class D WHERE D Resource The result of this query is depicted in Figure 10 21 References 1 4 5 S Abiteboul D Quass J McHugh J Widom and J Wiener The Lorel Query Language for Semistructured Data International Journal on Digital Libraries 1 1 68 88 April 1997 R G G Cattell D Barry Berler M J Eastman D Jordan C Russell O Schadow T Stanienda and F Velez The Object Database Standard ODMG 8 0 Morgan Kaufmann January 2000 V Christophides S Abiteboul S Cluet and M Scholl From Structured Documents to Novel Query Facilities In Proc of ACM SIGMOD Conf on Management of Data pages 313 324 Minneapolis Minnesota May 1994 V Christophides S Cluet and G Moerkotte Evaluating Queries with Generalized Path Expressions In
Download Pdf Manuals
Related Search