Modelling Student's Algebraic Knowledge with Dynamic Bayesian Networks

Abstract

Students' Knowledge Inference is an important component in the construction of Intelligent Tutoring Systems; it gets the basics of each learner knowledge, which allows the tutor to adapt the pedagogical instruction for each student. In the literature, it is common the use of Bayesian Networks to perform this kind of inference, because they are able to deal with uncertainties. This paper presents a Dynamic Bayesian Network modeling for the inference of student's algebraic knowledge. Differently from related works, it evaluates both student's procedural and declarative knowledge, besides being independent of problem. This paper also describes the steps we followed to get the information about the network probabilities, as well the evaluation conducted with the network. The evaluation results showed statistically significant similarities between the network inference and students' performance. This result evidences that the proposed work infers correctly student's knowledge.