ULI als Tesbed

1
A MINERVA based inference machine for metadate
course descriptions
Research seminar 09.07.2003 Learning Lab Rita
Gavriloaie, Jan Brase
2
Overview

• Metadata
• Annotating courses
• Inferencing
• MINERVA
• Demo

3
Metadata
4
Our ULI-subset of LOM
5
Inference rules - Motivation

• Inference rules provide default settings for
  metadata attributes. Goal Semi-automatic
  annotation of courses.
• Identifying the inference rules for attributes
  like hasPart, hasVersion, etc. we gain a formal
  definition of the different relationships between
  learning ressources.
• Furthermore we receive the implicit information
  out of the course description.

6
Extending the course description

• Using the Minerva inferncing machine on the
  rdf-file we create new descriptions
• ULI_Ki_extended.rdf
• ULI_InternetApplications_extended.rdf
• ULI_Algorithmentheorie_extended.rdf

7
MINERVA
8

• ISO-13211-1 Prolog compiler and executive hosted
  in Java, MINERVA is a commercial programming
  system for large private or public intelligent
  client-server applications on the internet
• In contrast to Java, MINERVA is declarative and
  high level. It includes a memory based database,
  search facilities, and pattern matching, all
  needed for intelligent interactive cooperating
  applications.

9
MINERVA RDF Parser

• Parse RDF statements rdf(S, P, O) into Prolog
  binary predicates P(S,O) and some ternary
  predicates for rdf containers (Seq ..)

10
Exporting Prolog to RDF

• One lecture metadata in one rdf file
• Algorithm
• find root resource
• find attributes of root resource (predefined
  attributes)
• export root resource and attributes
• forall subresources (depth first search following
  dcq_hasPart)
• find attributes
• export

11
Next stepBrowser applet
12
(No Transcript)
13
Demo