Architecture and algorithms of CPS - An agent-based tutoring system for coached problem solving on the Web

Currently, guiding students to reach the learning goals through problem-solving approach is quite primitive in most cyber-courseware systems on the Internet. In particular, these systems do not provide integrated learning-domain solutions for responding student enquiries based on individual student's background and interests. In order to offer student-specific problem-solving tutorship, dynamic generation of case-based solutions is a positive step towards this objective, which can accommodate individual student's needs accurately at anytime anywhere. In this paper, we present architecture and algorithms for developing an agent-based tutoring system called Coached Problem Solver (CPS) which, among other features, supports dynamic generation of student-oriented solutions on the Web. The facilities and capabilities of user profiles, rules, schema, XML and SMIL are incorporated and utilized by CPS so as to provide a better support to the interactive learning environment in which student-specific, content-based learning styles can be adequately entertained. A prototype system of CPS manifesting these algorithms has been built in Java, using IE on the PC platform.