A conceptual model for game based intelligent tutoring systems

Abstract

In order to build intelligent tutoring agents within games-based learning environments, practitioners must understand the three conceptual models used within Intelligent Tutoring Systems (ITS): the expert or domain model, the student model, and the instructional model. This paper investigates the inter-relationship between these models and how they combine to provide the expected behaviour of intelligent tutoring agents by representing and managing domain knowledge, applying techniques for monitoring the progress of human learning, and the appropriate selection of instructional strategies for individualised tuition. From understanding the application of these concepts, this paper proposes bi-directional human and machine learning as necessary for effective game-based intelligent tutoring systems. A conceptual architecture for game-based learning ITS implementations using multi-agent and machine learning technologies is then outlined, and the design of Stunt Robot a system for learning Newtonian Physics concepts is presented as a case-study.