Updating Student Model using Bayesian Network and Item Response Theory

Abstract

Nowadays different approaches are coming forth to tutor students using computers. In this paper, a computer based intelligent tutoring system (ITS) is presented. It projects out a new approach dealing with diagnosis in student modeling which emphasizes on Bayesian Networks (for decision making) and Item Response Theory (for adaptive question selection). The advantage of such an approach through Bayesian Networks (Formal framework of Uncertainty) is that this structural model allows substantial simplification when specifying parameters (Conditional Probabilities) which measures student ability at different levels of granularity. In addition, the probabilistic student model is proved to be quicker, more accurate and more efficient. Since most of the tutoring systems are static HTML web pages of class textbooks, our intelligent system can help a student navigate through online course materials and recommended learning goals.