Designing and implementing a rdfcoursefile maker

1
Designing and implementing a rdf-course-file maker
2
Motivation

• Annotating learning ressources with metadata is
  necessary.
• As a learning ressource is usually part of a
  course, there is a need for rdf-files for
  complete courses.
• Most part of this file(mainly the structure of
  the course) shall be created automatically by a
  tool

3
main issues and first ideas

• Analyse a representative number of course
  web-pages to understand what information can be
  filtered automatically.
• Write a formal description of LOM in TRILPE to
  establish a well-formed system of attributes with
  inference-rules (parallel to Research seminar)

4
Automatic Analysation of the structure of a course

• The locations of the learning ressources
  represent the structure of the course
• /chapter1/abcd.pdf
• /chapter1/abef.pdf
• /chapter2/sffe.pdf
• ....
• The ressources are stored in one folder, but the
  names of the files represent the structure
• /course1part1a.pdf
• /course1part1b.pdf
• /course1part2a.pdf
• ....
• This information could be used for an
  automatically suggested first structure of the
  course

5
Use of Tools

• Make use of existing open-source annotation and
  crawling tools
• Extend them with structure analysation and
  inference-rules for correct metadata suggestions
• The use the Jena-Toolkit to build the rdf-file
  perhaps?

6
State of the art

• http//km.aifb.uni-karlsruhe.de/kaon/Members/rvo/r
  df_crawler
• Rdf-crawler as part of the KAON-project - crawls
  web pages for rdf metadata information
• Document Classification and interpretation
  through the inference of logic-based
  modelsSemeraro, Ferilli Proceedings, ECDL 2001
• interesting steps in this direction, using
  INTHELEX (INcremental THEory Learner from
  EXamples)
• http//peccatte.karefil.com/Software/Catalogue/Cat
  alogueENG.htm
• Catalogue commercial windows application,
  collects standard metadata and transforms the in
  DC rdf.
• The idea of crawling html-pages for information
  is very common in the database community, in fact
  parsing of html files is quite simple.
• But none of these projects dealed with building a
  rdf-schema out of this information.

7
What to do?

• Analyse course decriptions to find a standard
  schema.(running)
• Write set of inference rules for metadata (almost
  done)
• Test existing html-page parser.
• Possible cooperation with J.Tane Karlsruhe, in
  their latest version of the rdf-crawler