SERAC : A Multi-agent System for the Evaluation and Revision of Knowledge in ITS

Abstract

Conceiving an automatic evaluation method loaded in a ITS is the first objective of this paper. This method relies on Reviser Agent that permits knowledge re-engineering from a multi-criteria evaluation (computer, ergonomic...). The second objective is to realise a general approach; so the revision mechanism has to be independent from the evaluation agents that transmit the knowledge. To model this method, we followed three steps. In the first stage, we designed an operational ITS (experimental case) with the help of an object oriented method (from Shlaer and Mellor) and object programming environment (Level5). In the second step, we realised an evaluation of this ITS by human experts to know the malfunctions. In this goal, we used a user task analysis (GOMS method) and an evaluation criteria grid. In the last step, obtained knowledge permits to conceive our automatic evaluation method based on a Reviser Agent (with Scheme and Level5).