An Adaptive Domain-Independent Agents-Based Tutor for Web-Based Supplemental Learning Environments

The Physics Interactive Video Tutor (PIVoT) is a Web-based multimedia resource for college-level Newtonian mechanics. The Personal Tutor (PT) is an intelligent tutoring system (ITS) integrated into PIVoT, assisting students and teachers in navigating through, understanding, and assessing PIVoT's educational media. PT is adaptive in that it personalizes its functionality to the preferences of its user. The combined PIVoT / PT system was designed to be domain-independent with respect to the style of pedagogy, models of user learning, and instructional algorithms. Thus, this design is easily adapted for use beyond the tested domain of introductory college physics. PT is designed in the object-oriented paradigm, building upon the recent work in multiagent systems (MAS). This agents-based approach, along with innovations in negotiating student-agent control and communication, allow current and future competing pedagogical strategies and cognitive theories to coexist harmoniously. New efficient, domain-independent techniques for discovering, updating, and presenting students' contextual interests improve information retrieval and site navigation. Unlike other computer-based instruction systems used as a tool for primary learning and assessment, PIVoT is used as a supplementary resource focusing on providing formative assessment to both student and educator alike. The PIVoT / PT system leverages reusability and system independence, two often-overlooked strengths of agent-based approaches to intelligent tutoring systems. Combined, PIVoT and the Personal Tutor provide an effective proving ground for innovations in intelligent tutoring system design that also reduces the cost of making such software. Thesis Supervisor: Steven R. Lerman Title: Class of 1922 Professor of Civil and Environmental Engineering This page intentionally left blank