NAT*LAB is a project that intends to exploit the most advanced technologies and know-how available today in Europe in order to start a cooperative, coordinated and long term effort addressing the issue of a methodology for (semi-)automatic knowledge acquisition for student modeling.
The technical kernel of the project consists of the design development, delivery, evaluation of incrementally more refined prototype L.E.S.s. From an initial "in vitro" prototype, by means of "inverted dialogues" (dialogues such that the L.E.S. plays the role of a student and the student "debugs" the system) one collects a growing catalogue of possible interpretations of mistakes (misconceptions) that enrich the subsequent versions of the system, when they have been included into the knowledge available to the L.E.S. Insofar as this inclusion will be automatic, we will be able to talk of a kind of "learning" by the L.E.S. as a result of its own experience with the student. The project includes specific contributions to the methodology from workpackages that will investigate issues in machine learning, second generation expert systems, dialogue management, reason maintenance in a multiple belief space. As the software / hardware technology needed for obtaining a reasonable efficiency with the before mentioned tools is currently above the one available for the delivery of courseware, we will also evaluate how concurrent machines, at present just appearing on the market, may in the medium term be adequate as implementation media with respect to reaction times and costs of the resource.
Società per la Didattica con l'Elaboratore Dida El Srl – ITALY, VRIJE UNIVERSITEIT BRUSSEL – BELGIUM, HITEC LTD – GREECE, CINECA – ITALY, UNIVERSITY OF LEEDS - UNITED KINGDOM, ICL - STC - UNITED KINGDOM
programma CEE/framework: DELTA, 2PQ