Feedback provision strategies in intelligent tutoring systems based on clustered solution spaces

Abstract

Designing an Intelligent Tutoring System (ITS) usually requires precise models of the underlying domain, as well as of how a human tutor would respond to student mistakes. As such, the applicability of ITSs is typically restricted to welldefined domains where such a formalization is possible. The extension of ITSs to ill-defined domains constitutes a challenge. In this paper, we propose the provision of feedback based on solution spaces which are automatically clustered by machine learning techniques operating on sets of student solutions. We validated our approach in an expert evaluation with a data set from a programming course. The evaluation confirmed the feasibility of the proposed feedback provision strategies.