The University of Edinburgh -
Division of Informatics
Forrest Hill & 80 South Bridge

PhD Thesis #9603

Title:The Use of Belief Revision to Model Contradictions and Entrenched Misconceptions
Date: 1996
Abstract:The goal of this research is to illustrate the potential of belief revision in a tutoring system. The application is in a domain where contradictions arise when learners commit errors, and where misconceptions leading to these errors are deeply entrenched in the learners' reasoning process. We show how a belief revision system which employs the notion of epistemic entrenchment in resolving conflict can be useful in modelling learners and for remediation. Our particular focuses on: 1. the domain of translating simple relational algebra word problems to algebraic equations; 2. misconceptions related to the reversa1 error i.e formulation of a reversed equation; 3. learner being persistent in holding the misconceptions which are referred to as the learner's entrenched misconceptions. Our work also offers a possible learner modelling technique for contradictions and entrenched misconceptions - diagnosing through contradiction of beliefs and assisting the remediation process by using the beliefs and justifications involved when resolving the contradictions diagnosed. We illustrate such investigations in TRAPS (Translating Relational Algebraic Problems System). TRAPS, a prototype computer-based tutoring system has been developed with a belief revision system embedded in its learner model. The belief revision system assists TRAPS in modelling and mending a learner's misconceptions. It also helps TRAPS in making overall observations of the learner's performance. ii

[Search These Pages] [DAI Home Page] [Comment]