Towards New Learning Strategies in Intelligent Tutoring Systems

Various co-operative tutoring systems have been recently proposed as a continuation of intelligent tutoring systems. They encourage learning as a jointly constructed knowledge rather than a transfer of expertise from the teacher to the learner. These environments can be strengthened using elicitation techniques which encourage the acquisition process from a learner point of view. In this article we present the different forms of co-operative systems and the advantages of elicitation techniques that can also handle the context in which the knowledge should be transferred. We detail three new forms respectively called learning by disturbing, teaching by co-operation, learning by over-teaching, and the corresponding algorithms of co-operative systems based on elicitation. Finally, we present the architecture of a multi-co-operative system that uses elicitation techniques together with multiple strategies of co-operation.