TOWARDS A LEARNING MODEL BASED ON BAYESIAN NETWORKS

Under uncertainty Bayesian networks are effective tools for learner modeling. They have been successfully used in many systems, with different objectives, from the assessment of knowledge of the learner into the recognition of the plan followed in problem solving. Many models have been constructed, but there has been no attempt to synthetic approach to the problem. Our goal is to study some methods of implementing a Bayesian network. Our work focuses on the question of the orientation of the arcs, and more generally on the structure of Bayesian network modeling of the learner. We try to show in this work how this question is crucial. In addition, the issue of structural adjustment in the network behavior of the learner sometimes had been suggested, and while different results from cognitive psychology attests to the existence of structural differences by level of expertise. The central hypothesis of our work is that has been a link between the structure of the learner model and level of expertise. We present our probabilistic graphical models of multi-networks to take into account several networks within the same model. The experiments presented in this work are arguments in favor of our hypothesis on the link between the level of expertise of the learner and the structure of Bayesian network.