Database Creation and Information Extraction from
Contents
1. Alternate Location Notes Examples Tag lt province_state gt Description Standard Location Home state province of the author if listed Biography Alternate Location Notes Examples 3 2 DATABASE The database provided was made in Microsoft Access 2013 and should be very simple to use The database will also open with no problems on a Microsoft Access 2010 installation although previous versions were not checked Once the database is open double click on the etd revised entry under the Tables tab on the far left side of the screen You will find the attributes listed across the top of the table starting with num_pages To edit an existing entry scroll to find the attribute for the entry that you want to change and click anywhere in that cell To add a completely new entry scroll to the bottom of the entries to a row marked with an asterisk on the far left and begin entering information for each attribute in this row You will notice that the asterisk changes to a pencil icon and a new row is created below with the asterisk next to it To delete an existing entry click on the far left box in the row you want to delete in order to select it then right click and select Delete Record In order to export the database as a formatted XML file click on the External Data tab at the top in the same menu as file There will be three p2. lt acknowledged funding org 3 gt Only about a fifth of the recorded ETDs authors explicitly mention the source of the funding The rationale for tracking this information is that if the author felt their funding sources to be worth noting it may be important in their decision to begin research lt prior_workplace_1 gt lt prior_workplace_2 gt lt prior_workplace_3 gt The author s industry experience may influence what kind of research is done or how it is conducted Analysis could be conducted to see which jobs move back into research more than others if any among other things lt prior_research_1 gt lt prior_research_2 gt Previous research experience is very likely to lead to more research and analysis of this attribute could provide information about changes in research habits with age or other factors or how research interests are distributed across the population of ETD authors lt country gt lt city gt lt province_state gt This information is useful to track where the author comes from lending itself to analysis on research done by students of different national origin 4 3 STRUCTURE OF THE DATABASE The database currently consists of 2 tables lt etd revised gt and lt Research Fields gt lt etd revised gt is the primary table that contains the stored ETD data lt Research Fields gt contains a listing of the various research categories used to classify documents in this project There is a one
3. lt advisor gt Description Standard Location The primary advisor of the author Acknowledgements Alternate Location Cover Page Notes Examples Nearly all of the authors in the ETDs filed for this project make an explicit reference to their advisor All name formatting is retained In the case of multiple advisors the first name listed is tagged If there is no acknowledgments section or advisor mentioned by the author this tag is left blank Tag lt advisor gt Description Standard Location The primary advisor of the author Acknowledgements Alternate Location Cover Page Notes Examples Nearly all of the authors in the ETDs filed for this project make an explicit reference to their advisor All name formatting is retained In the case of multiple advisors the first name listed is tagged If there is no acknowledgments section or advisor mentioned by the author this tag is left blank Tag lt num_acknowledged_colleagues gt Description Standard Location Records the number of explicit references to professors including the advisors fellow students and colleagues made by the Acknowledgements Alternate Location Biography Notes Examples author Persons are only counted once no duplicates Tag lt num_acknowledged_friends gt Description Standard
4. Location Records the number of nun professional explicit references made to friends and family of the author The references counted for this tag should not overlap with those in lt num_acknowledged_colleagues gt Acknowledgements Alternate Location Biography Notes Examples Group terms such as family or friends should not be counted unless there are no references to specific persons Tag lt undergraduate_institute_name gt Description Standard Location Name of the author s undergraduate institution Biography Alternate Location Notes Examples Tag lt undergraduate_institute_degree gt Description Standard Location Name of the author s undergraduate degree Biography Alternate Location Notes Examples Tag lt acknowledge funding org _1 gt lt acknowledge funding org 2 gt lt acknowledge funding org 3 gt Description Standard Location Names of any explicitly mentioned funding organizations Biography Alternate Location Cover Page Abstract Notes Examples Numerical suffixes should be dropped and the general name of the funding body retained Example US Department of Energy Grant No 005 D341 gt US Department of Energy The tags are filled in the order in which the funding organizations are encountered th
5. most ETDs is similar and is normally in this approximate order dependent on institution of origin title page table of contents abstract actual content biography acknowledgements and resume not normally present In these all but the table of contents and the paper itself contains required information for the database The instructions provide the most common locations for each tag attribute and alternate locations if any were found They also instruct the Mechanical Turk user on what to do in case of missing data for each attribute 3 USER MANUAL 3 1 AMAZON MECHANICAL TURK PROCEDURE This section assumes that the user is familiar with Amazon Mechanical Turk and is using the following instructions to extract data from ETDs by hand For each tag the user should find the information in the standard location or the alternate location s if necessary All notes and examples should be followed and referred to Unless otherwise stated if an attribute cannot be found within the listed standard or alternate locations the attribute should be left blank Table 1 ETD Tag Descriptions Tag lt title gt Description Standard Location Records the title of the ETD Cover Page Alternate Location Notes Examples Tag lt research _category gt Description Standard Location An integer ID representing one of the fields of computer Extrapolated science research that best categorizes this pap
6. to many relationship between lt etd revised gt and lt Research Fields gt where an entry in lt etd revised gt can have one of many lt Research Field gt categories applied to it The structure of this database was kept purposefully simple to avoid imposing any classification restrictions For example there are currently many duplicate entries under the lt institution gt attribute of the database To fix this the name of a university should be standardized or a mapping table made The current database platform Microsoft Access 2013 includes features for creating simple front end user interfaces for editing data This is not an essential portion of the database and will likely not transfer to another platform This project s interface a portion of which is depicted below in Figure 4 contains simple text and drop down elements for editing data 13 ETD Entry Form ETD_Title Factors Affecting the Programming Performance of Computer Science Students Research_Field Education Author John B Raley Institution Virginia Polytechnic Institute and State University submission_year Advisor Dr Cliff Shaffer Num_Acknowledged_ Colleagues Num_Acknowledged_Family 3 Undergraduate_Institute_Name Mathematics Undergraduate_Degree Figure 4 Microsoft Access Record Entry Form 14 5 LESSONS LEARNED 5 1 TIMELINE o 2 22 13 B
7. Database Creation and Information Extraction from ETDs CS 4624 Virginia Tech Blacksburg VA Spring 2013 Client Venkat Srinivasan By Lamont Banks Joseph Luke 1 CONTENTS 2 AOS MAC E hese eases anebledasetnees wikia ces aed seas toe ede see ate eed a ee ae eee 3 3 USer Manuals cities ee nas A hia a eg A ee ae nae 4 3 1 Amazon Mechanical Turk Procedure esceeecceesseceesceceeeeeceaeeeeaaeceeneeceaeeeeaaeeeeaaeseeeesaeeneaaeeeeaeeee 4 3 2 Dita ui A baa eae eed 9 4 Developers Mandaly recort eeen chad A dit dp ees 10 4 1 XML T g Description h aa a A ees teed A aA A A added 10 4 2 Rationale for XML AES e ee AS 11 4 3 Structure of th Database icre oalcedesede dade Ginsbavsste dua edoucseessladsds dbaadsepacadvis dale dde dada dona inaa eei 13 A A O 15 5 1 Timeline dadas 15 5 2 Problems ii A 15 5 3 Future iMpProvemMentS ieres Esa 16 6 AcknoWledeementse dr de dec ada deb de ed do dd ra 17 To Referentes sesonkia e a cane aad aaa a aa aaa a dh ae hase eae 18 2 ABSTRACT This project was done at the behest of the Computing Research Association CRA The main point of the project was to collect data associated with electronic theses and dissertations ETDs to allow determination of why graduate students in computing go into computing research The deliverables include a database of the ETDs analyzed and a framework for manual approaches to this data extraction To accomplish these objectives ETDs from Nor
8. anels in the new toolbar the one on the left starts with Saved Imports and the one next to it on the right starts with Saved Exports Click on XML File in this second panel and follow the prompts to indicate location and name of the file and what should be exported data schema presentation Leave the default option which is to export data and schema first two check boxes selected unless you know that you need the third option Also included in the database file is another table entitled Research Fields This can be edited in the same manner to add remove or modify existing research fields to use in the tag Research_Field 4 DEVELOPER S MANUAL 4 1 XML TAG DESCRIPTION There were several XML tagging structures that were developed and tested throughout the project The technical descriptions of the current tagging structure are described below 1 Software Engineering 2 High Performance Computing 3 Media and Visualization 5 Human Computer Interaction 6 Education 7 Data Management and Information 8 Networking 10 Biologial Computing 11 Theoretical Research 12 Security 13 Artificial Intelligence Figure 1 Research Fields 10 title gt Analyzing Software Artifacts Through Singular Value Decomposition to Guide Development Decisions lt title gt Primary Category gt 1 lt Primary Category gt Author gt Mark Stephen Sherriff lt Author gt Institution gt North Carolina State University lt I
9. egan reading over resources started to analyze and document ETDs looked for information about the author and where that information was in the paper o 3 8 13 Started to draft techniques to use for analysis and outlined database with XML tags o 3 22 13 Developed preliminary database and began to develop Amazon Mechanical Turk procedures o 3 25 13 Presented midterm presentation o 4 19 13 Finished Amazon Mechanical Turk procedures regarding how to find useful information in an ETD o 5 3 13 Finished database and created final presentation o 5 6 13 Presented final presentation 5 2 PROBLEMS Most of the problems experienced with this project had to do with missing or incomplete information Most ETDs did not have all of the information predicted in our XML tags and it was decided to leave these blank rather than try to extrapolate this information from the rest of the paper except as where mentioned To prevent confusion and differences in data entry only explicit mentions of a particular attribute were entered into the database When faced with multiple locations for an attribute author s name in multiple places for example the most explicit location was used this was normally the first mention Another small issue that was encountered was page numbering The system created provides an attribute for number of pages num_pages but most ETDs have pages numbered both with Roman numerals first tit
10. er Alternate Location Notes Examples The categories and their respective IDs are listed in Figure 1 Left to the cataloguer s discretion Tag lt author gt Description Standard Location Cover Page Records the author of the paper Alternate Location Notes Examples All formatting included Tag lt institution gt Description Standard Location Cover Page Alternate Location Name of the school for which the ETD is filed Notes Examples Currently all variations on institution names are retained Example Virginia Tech is recorded as well as the longer Virginia Polytechnic Institute and State University Tag lt submission_year gt Description Standard Location Year the ETD was submitted for review Cover Page Alternate Location Notes Examples Tag lt advisor gt Description Standard Location The primary advisor of the author Acknowledgements Alternate Location Cover Page Notes Examples Nearly all of the authors in the ETDs filed for this project make an explicit reference to their advisor All name formatting is retained In the case of multiple advisors the first name listed is tagged If there is no acknowledgments section or advisor mentioned by the author this tag is left blank Tag
11. ere is no tagging hierarchy Tag lt prior_workplace_1 gt lt prior_workplace_2 gt lt prior_workplace_3 gt Description Standard Location Names of any organization the author explicitly states having previous experience Biography Alternate Location Acknowledgements Notes Examples in after completing their undergraduate degree Our schema currently does not record the author s title position while working Tag lt prior_research_area_1 gt lt prior_research_area_2 gt Description Standard Location Any stated research expertise of the author before entering graduate school Biography Alternate Location Notes Examples Example excerpt from an ETD biography spent her senior year working for Dr Michael Young on Liquid Narrative projects continued to work for LNG in pursuit of a Master s degree through the summer of 2001 The research area of the Liquid Narrative projects Media and Visualization in this case would be recorded in the tags Tag lt country gt Description Standard Location Home country of the author if listed Biography Alternate Location Notes Examples Do not assume the country if not listed Tag lt city gt Description Standard Location Home city of the author if listed Biography
12. le page through abstract and then with standard Arabic numerals for the rest of the paper To solve this initially the total number of pages was used regardless of numbering Any page numbers used in the system were then indexed off of the title page or first page if different being page 1 and ignoring the nominal system However page numbers were removed from the XML schema later on in the project and this ended up not affecting the final project Although not critical problem another issue was deciding which universities and which papers from those universities to use In the end convenience and access to web sites restricted the universities to 15 North Carolina State University NCSU Florida State University FSU Auburn University AU Wake Forest University WFU and Virginia Tech VT Papers were then selected randomly from those available However this is an aspect of the project that can be improved in the future 5 3 FUTURE IMPROVEMENTS This project could be improved in many ways both in its construction and its usage The most obvious improvement is simply to add more entries The more data that is contained in the database the better any analysis of that data will be Secondly the database structure could be expanded to better use many of Access s features cross relations This would allow for more use cases querying for various combinations of attributes but would restrict the database to Access Similarly
13. nstitution gt submission year gt 2007 lt submission year gt Advisor gt Dr Laurie Williams lt Advisor gt Num_Acknowledged_Colleagues gt 22 lt Num_Acknowledged_Colleagues gt Undergraduate Institute Name gt Wake Forest University lt Undergraduate Institute _Name gt Undergraduate Institute _Degree gt Computer Science lt Undergraduate Institute Degree gt Acknowledged Funding Org 1 gt IBM lt Acknowledged Funding Org 1 gt Acknowledged Funding Org 2 gt Center for Advanced Computing and Communication lt Acknowledged Funding Org 2 gt Acknowledged Funding Org 3 gt National Science Foundation lt Acknowledged Funding Org _3 gt Prior Workplace 1 gt IBM lt Prior Workplace_1 gt Prior Research Area 1 gt 1 lt Prior Research Area 1 gt Num_Acknowledged_Family gt 7 lt Num_Acknowledged_Family gt Country gt United States lt Country gt City gt Salisbury lt City gt Province State gt North Carolina lt Province State gt etd gt Figure 2 Sample Completed XML Structure 4 2 RATIONALE FOR XML TAGS The overall goal of the XML structure was to capture information that would be useful in determining why students enter graduate research The following is the description of the tags used lt title gt lt author gt lt institute gt lt submission_year gt The ETD database should contain basic information about the dissertation these tags were created to track these basic details lt research_category gt Data mining of
14. order but the search parameters were varied while tagging documents Meaning ETDs were added after sorting alphabetically sorted by year or by author s name A more optimal selection method would should be used to get the most value out of this tag lt advisor gt Any algorithms tracking this tag may find links between specific professors within a given university and the inspiration for the author s move into research For example an analysis of the ETDs may find that a particular professor is frequently cited as the motivation for students going into research or perhaps a large proportion of professors from a particular university is responsible for graduate student enrollment lt num_acknowledged_colleagues gt lt num_acknowledged_friends gt The intent behind this tag was to help identify the prevalence of presumably strong networks of supports in completing research projects lt undergraduate_institute_name gt lt undergraduate_institute_degree gt The schools that send significant numbers of students into graduate schools could possibly be investigated and studies to determine what programs and curriculum they have implemented to achieve their results The degree of an overwhelming majority of the catalogued ETDs were in computer science or extremely similar fields but this cannot be assumed for all students so this data has been tagged 12 lt acknowledged funding org 1 gt lt acknowledged_ funding org 2 gt
15. reless Networks with Joint Scheduling and Congestion Control Nimbhorkar Sankalp 5 Sep 2001 Adaptive Multipath Traffic Allocation in TCP IP Networks Elhaddad Mahmoud Shawky 28 Mar 2000 An Adaptive Non parametric Kernel Method for Classification Kaszycki Gregory John 20 Nov 2008 Adding Coordination to the Management of High End Storage Systems Zhang Zhe 18 Oct 2005 Address Space Layout Permutation Increasing Resistance to Memory Corruption Attacks Bookholt Christopher Gi Gien May 2011 Advanced Learning Techniques for Improved Inference of Bayesian Belief Networks from Uncertain and High Pansombut Tatdow dimensional Data 21 Aug 2002 Affective Behavior Control for Lifelike Pedagogical Agents Stelling Gary Dean 6 Dec 2011 ALACRI2TY Lossless Dats Compression for Analytics driven Query Processing Arkatkar Isha 22 Jul 2008 Algorithms for Selecting Views and Indexes to Answer Queries Kormilitsin Maxim 6 Sep 2002 Analysis and Quantification of Test Driven Development Approach George Boby 10 Dec 2012 Analytical Approach for Bot Ch Detection in a Massive Multiplayer Online Racing Game Villanes Arellano Andres 21 Jul 2008 Analytical Bounding Dsts Cache Behavior for Real Time Systems Ramaprazad Harini 25 Jan 2009 Analytical Models and Efficient Dimensioning Algorithms for Communication Systems In Randomly Changing Trafic Tian Wenhong Environments Figure 3 Sample Search Results page In this figure results in descending alphabetical
16. th Carolina State University NCSU Florida State University FSU Auburn University AU Wake Forest University WFU and Virginia Tech VT were analyzed and inserted into the database Extensible Markup Language XML was decided upon as the structuring format for the ETDs and a tag structure was created utilizing biographical educational and institutional data from each ETD Some of the tags included author name title of the paper year published undergraduate institution of the author etc XML was chosen because of its prevalence in the ETD field its structural properties and ease of use These tags were used to create the attributes for each entry in the database in Microsoft Access Access was chosen mostly because of convenience and easy porting of tags into the system However the database could be moved into another system quite easily In order to move the database it would be converted to XML and then imported into a MySQL database or Oracle Challenges that arose included missing data or insufficient information in various areas For instance many papers lacked information about source of funding country of origin and information about the author The second deliverable took the form of instructions pg 4 to an Amazon Mechanical Turk user on how to extract information These instructions were created and provided in order to increase speed and decrease errors in manual data extraction It was found that the basic structure of
17. the XML tags could be expanded to allow specific or just larger quantities of information from ETDs Lastly one piece of the project that could not be completed given the time requirements was to analyze the data Analysis with a program such as WEKA could provide many relationships and clusters that are not immediately obvious This would also allow testing predictions and applying machine learning algorithms to the data in order to predict attributes of future waves of computing researchers 16 6 ACKNOWLEDGEMENTS We would like to acknowledge Dr Edward Fox for helping us throughout this project and filling in as client when ours was unable to perform this function this semester We would also like to thank our original client Dr Venkat Srinivasan for the idea and impetus for the project 17 7 REFERENCES Aniket P amp Fox E 2002 XML for ETDs Department of Computer Science Virginia Polytechnic Institute and State University Blacksburg VA 18
18. the collected data could reveal interesting trends in what types of research students pursed Any useful data that data mining would yield is highly dependent on a balanced representation of the various fields For this project ETDs were selected randomly to the best of the compilers abilities Specifically a search for Computer Science related papers was done on various university ETD databases and 4 5 papers were chosen from each page of results Figure 3 shows a sample search results page 11 Sort by title lv In order Ascending Y Results Page 20 Y Authors Record Ally Update Showing results 1 to 20 of 482 next gt Issue Date Title Author s 26 Mar 2012 The 1 k Eulerisn Polynomials Viswanathan Gopal 28 Mar 2002 Abstraction Based Generation of Finite State Models from C Programs DuVamey Daniel C 7 Jul 2003 Abstraction Based Static Analysis of Buffer Overruns in C Programs Srinivasa Gopal Ranganatha 8 Jan 2008 Access Point Coordinated Contention Resolution for Channel Access in Wireless LANs Mahboob Ahmed Mishar 12 May 2010 Achieving Communication Scalability in Collaborative Development Tools Performance Modeling of the Jazz Phadnis Deepti Development Environment 22 Aug 2008 Active Timing Based Techniques for Attack Attribution through Stepping Stones Peng Pai 28 Aug 2002 ADaPT Adaptive Development and Prototyping Technique Srikanth Heme Lakshman 16 Mar 2012 Adaptive Channel Width Allocation for Multihop Wi
Download Pdf Manuals
Related Search