A Bayesian Network Model for More Natrual Intelligent Tutoring Systems

A key element in the design of Intelligent Tutoring Systems (ITS) is to fit the learning plan to the individual variances of learners. We studied this problem by modeling a dependency relationship between concepts of a Java programing course within a Bayesian network. Our model was established on the basis of experts’ views, corroborated by experimental investigation. We found that the tested programing concepts do, indeed, exhibit dependency relationships, which could minimize the learning time of the individuals, if they were taken into account. This, implies that for the design of a more natural ITS, it is helpful to consider the existent concept dependency, so as toenhance the role of individual differences between the learners and, hence, minimize the learning time.