Learning what is important to learn, some experiments with inductive logic programming

Some Experiments with Inductive Logic Programming F. Jacquenet M. Bernard C. Nicolini LIRSIA EURISE CEA Valduc Universit e de Bourgogne Universit e de Saint-Etienne 21011 Dijon Cedex 42023 Saint-Etienne Cedex 2 21120 Is/Tille France France France Abstract Most Intelligent Tutoring Systems (ITS) nowadays integrate some Arti cial Intelligence techniques to improve the quality of the Computer Aided Tutor. Knowledge bases, reasoning techniques, can be widely used. Nevertheless, the problem is \What is important to learn for a student". This question is an important bottleneck for the design of a good and e cient ITS. From the way we answer to it will depend the success of the ITS being designed. In this paper, we propose to integrate some Machine Learning techniques in ITS to allow them to automatically improve their knowledge bases and reasoning facilities.