ConnectA

1
ConnectA!

• Intelligent Search Engine Bases on Authors
  Connectivity
• Hamidreza Baghi
• Mohammadreza Barouni-Ebrahimi
• Reza Zafarani

2
Problem and Description
Harold Boley
Bruce Spencer
?
?
?
Semantic Web

• Documents of the related authors?
• Active authors in period 2000, 2006?
• Related authors in a specific type of documents
  (e.g. proceedings)?

3
Solution
Documents
Keyword Author(s)
Author Keywords
Application Engine
Author
Related Authors
Documents
OOjDREW
KB
4
Solution

• User asks for next level

Next Level
Application Engine
OOjDREW
KB
5
Demo Description
6
Demo Description
7
Facts

• The facts store information for papers
• Author
• Year
• Type (journal, conference)
• author("Anna Maclachlan","conf/iastedWTAS/Maclachl
  anB05").
• author("Harold Boley","conf/iastedWTAS/MaclachlanB
  05").
• date(2005Integer,"conf/iastedWTAS/MaclachlanB05")
  .
• type("inproceedings","conf/iastedWTAS/MaclachlanB0
  5").

8
DBLP XML

• The facts are extracted from the DBLP XML
  database
• DBLP to POSL converter is implemented using
  Xerces XML Parser

ltinproceedings mdate"2005-10-27"
key"conf/iastedWTAS/MaclachlanB05"gt ltauthorgtAnna
Maclachlanlt/authorgt ltauthorgtHarold
Boleylt/authorgt lttitlegtSemantic Web Rules for
Business Information.lt/titlegt ltpagesgt146-153lt/page
sgt ltyeargt2005lt/yeargt ltcrossrefgtconf/iastedWTAS/200
5lt/crossrefgt ltbooktitlegtWeb Technologies,
Applications, and Serviceslt/booktitlegt lturlgtdb/con
f/iastedWTAS/wtas2005.htmlMaclachlanB05lt/urlgt lt/i
nproceedingsgt
9
DBLP Structure
DBLP
article
inproceedings
title
author
year
author

key(att)
title
author
year
author

key(att)
10
DBLP DTD

• lt!ELEMENT dblp (articleinproceedingsproceedings
  bookincollection
• phdthesismastersthesiswww)gt
• lt!ENTITY field "authoreditortitlebooktitlepa
  gesyearaddressjournalvolumenumbermonthurle
  ecdromcitepublishernotecrossrefisbnseriess
  choolchapter"gt
• lt!ELEMENT article (field)gt
• lt!ATTLIST article
• key CDATA REQUIRED
• reviewid CDATA IMPLIED
• rating CDATA IMPLIED
• mdate CDATA IMPLIED
• gt
• lt!ELEMENT inproceedings (field)gt
• lt!ATTLIST inproceedings key CDATA REQUIRED
• mdate CDATA IMPLIED
• gt

11
Rules and Query

• Recursive rules are used to find the related
  authors
• Using built-ins
• Limit the derivation tree to get the related
  authors up to expected level.
• Query
• related(Harold Boley , ?x).
• related_date(Harold Boley , ?x , 2000Integer ,
  2005Integer).
• related_type(Harold Boley,?x, article).

12
GUI
The Java Program Creates the status and HTML
(once the work is done)
The PHP UI runs the java program in back ground
The AJAX checks the status and retrieves the HTML
when the status shows that the HTML is created
JAVA Program
HTML
STATUS
13
Future Work

• Optimizing the system to handle a large
  knowledge base.
• Extending the system by having more constraints.
• Limiting the graph search space based on
  authors interest.

14
Thank you! Questions?