A team of agents cooperating for intelligent tutoring

This paper presents how the tutoring knowledge in an ITS, in particular in a domain involving problem-solving skills in addition to conventional domain theory, can be designed and organised as a team of cooperating intelligent agents. After a brief description of the domain knowledge and a generic functioning framework for ITSs in problem-solving domains, we introduce how the tutoring knowledge is modelled as pedagogical agents. We then show how these agents are designed to interact with one another and with the student, and discuss about techniques which may help an agent to decide to intervene in the process in order to fulfil a given pedagogical function, henceforth making pedagogical interactions smoother and hopefully more efficient. To illustrate our point of view, we then present and comment, in a particular application domain, a problem-solving exercise and a possible tutoring dialogue based on that exercise. Finally, referring to that example, we show a few techniques by which such a dialogue can be achieved. Such a cooperative-agent-oriented model should both facilitate the implementation of the tutoring knowledge in an ITS and help improve the student's comprehension through more efficient pedagogical interactions.