A Proposal for a New Framework of Learner Modeling: From Modeling “Understanding” to “Not Understanding”

Abstract

In the study of intelligent tutoring systems (ITSs), research on learner modeling has been actively conducted to provide effective support adapted to a learner's state of understanding, in which learning history, learning style, lack of knowledge, misunderstandings (bugs), etc. have been discussed. On the other hand, low-performance learners who genuinely need assistance find themselves in trouble because they do not know what to do, that is, they are in a state of “not understanding.” As far as the author knows, however, there is no research on modeling the state of “not understanding.” In this study, we propose a new methodology of modeling the state of “not understanding” of a learner directly based on the state space and search model used in the General Problem Solver (GPS) in AI. Then, we develop a theory for generating feedback that adequately supports low-performance learners, which can be justified by the theory of the proposed learner model.