Designing intelligent tutoring systems: a multiagent Planning Approach

Intelligent Tutoring Systems (ITS) are computer aided intelligent learning tools. Most recent architectures of these systems have focussed on the tutor or curriculum components, but with little attention being paid to planning and intelligent collaboration between the different components. In this paper, we propose several improvements by describing a new architecture that involves sophisticated planning processes at different levels of the ITS processing and by decomposing the tutor into two different components, one specialized in the tutorial actions planning and the other tailored for the generation of multimedia presentations. Moreover, the user interface is made more robust and flexible thanks in part to the use of planning techniques. These improvements are built on Nkambou, Gauthier and Lefebvre's architecture, but we believe that the most of the important ideas discussed herein may also be exploited in other architectures.