Deductive Knowledge Management for Personalized Documents

Peter Baumgartner, Ingo Dahn


This work is embedded in the TRIAL-SOLUTION project, which is funded by the EU as part of its Information Society Technologies Programme (IST) within the EU's Fifth RTD Framework Programme. The project consortium consists of 12 partners from five European countries, and is coordinated by the Universität Koblenz.

The TRIAL-SOLUTION project aims to develop a technology for the generation of {\em personalized teaching materials} - notably in the field of mathematics - from existing documents. The background of this project is the "Slicing Book Technology", which has been developed by the second author. The readily developed tools enable authors, teachers and students to produce personalised teaching or learning materials. With these tools, a mathematical text book ws prepared and is distributed as a commercial product [Wolter:Dahn:AnalysisIndividuell:Springer:00]. The TRIAL-SOLUTION project will extend the Slicing Book Technology by possibilities to generate personalized documents from various sources.

At the heart of this enterprise are techniques to handle the knowledge coming from the various sources. These sources include (i) different books that are sliced into a number of small objects, such as theorems, proofs, and so on, (ii) a knowledge base of metadata on content (e.g. by keywords), didactic features, and interoperability interfacing, (iii) the user profile, including e.g.\ information about units known to him, and (iv) thesauri that help to categorize and connect knowledge across different books.

All these sources are to be taken into account when generating a personalized document. A typical query would be, for instance, "Assemble from all books in the repository all sections on `cartesian product' and related concepts, but exclude those sections I know of, and exclude exercises.".

We have begun modelling the mentioned sources with first-order predicate calculus. Further, we are currently experimenting with and adapting our FDPLL theorem prover [Baumgartner:FDPLL:CADE:00], in order to answer queries of the mentioned form. In the talk, we will report on the current state of affairs.


Christoph Benzmüller, chris@ags.uni-sb.de
Last modified: Wed Sep 13 19:56:02 MEST 2000