Inferencing over RDF Metadata course descriptions

1
Inferencing over RDF Metadata course descriptions

• Research seminar WS 2002/2003

2
Melting the projects

• The work on course descriptions in this seminar
  is now combined with the search for inference
  rules for LOM metadata (Together with M. Painter)
  and the implementing of a Edutella-RDQL-Query
  peer in the context of the ULI project (Together
  with U. Thaden) To be presented on the 14. Of
  march at the next ULI meeting.

3
Schedule for the next weeks

•  
• February 6th week
• -         finishing with inference rules
  representation in PROLOG.
• -         Work on Minerva.
•  
• February 8th week
• -         Tool for completing ULI course
  description with Minerva, PROLOG inference
  rules in context of Edutella/RDQL interface.
•  
• February, March 10th week
• -         first attempts of crawling course
  pages.
• -         cooperative meeting with J. Tane
  (Karlsruhe) on their lates version of their
  rdf-crawler.

4
6th weekfinishing with inference rules
representation in PROLOG.

• A first test-version of the rule set has been
  implemented
• Covering for example rules like
• dc_creator(Resource1,Object)-
• dcterms_hasPart(Resource1,Resource2),
• dc_creator(Resource2,Object).
• dcterms_requires(Resource1,Object)-
• dcterms_hasPart(Resource1,Resource2),
• dcterms_requires(Resource2,Resource3),
• not dcterms_hasPart(Resource1,Resource3).

5
Test course description

• The rules are tested on the latest description of
  the Artificial intelligence course
• dc_title('Intelligenz','Uli Kuenstliche
  Intelligenz WS 2002 (Hannover)').
• dc_description('Intelligenz','Einführung in die
  Grundprinzipien der Künstlichen Intelligenz
  Semantische Netze und Suchalgorithmen, Regeln und
  Regelverkettung, Frames und
• Vererbung, Constraint Propagation, Logik und
  Resolution, PROLOG und Maschinelles Lernen').
• dc_creator('Intelligenz','Wolfgang Nejdl').
• dcterms_created('Intelligenz','2002-09-15').
• dcterms_hasPart('Intelligenz','Modul1').
• dcterms_hasPart('Intelligenz','Modul2').
• dcterms_hasPart('Intelligenz','Modul3').
• dcterms_hasPart('Intelligenz','Modul4').
• dcterms_hasPart('Intelligenz','Modul5').
• dcterms_hasPart('Intelligenz','Modul6').

6
Still to do

• Getting it together
• RDF file
• Minerva
• PROLOG inference system
• Edutella-RDQL Interface
• Creating RDF files with Minerva

7
Problems

• The import/export RDF-file ? PROLOG-Knowledge
  base ? RDF-file is a bit awkward.
• Inferenced new metadata descriptions are
  Asserted in PROLOG. Including them in the
  RDF-File needs more information than just the
  simple fact.