MathTutor: A Multi-Agent Intelligent Tutoring System

In this paper we propose a multi-agent intelligent tutoring system building tool that integrates different formalisms in order to facilitate the teacher task of developing the contents of a tutorial system and at the same time to provide adaptiveness and flexibility in the presentation. The adopted formalisms are ground logic terms for the student model, data-bases for the domain model and object Petri nets for the pedagogical model. The interaction between the student and each agent of the system is controlled by an object Petri net, automatically translated into a rule-based expert system. The object Petri net tokens are composed by data objects that contain pointers to the student model and to the domain knowledge, stored into a data-base of texts, examples and exercises. The object Petri net transitions are controlled by logical conditions that refer to the student model and the firing of these transitions produce actions that update this student model.