Why should an ITS bother with students’ explanations?

Abstract

PROBIT (PROBability Intelligent Tutor) served as a test bed for examining the potential roles of students’ explanations in an ITS. PROBIT’s environment was modified to enable the student to use natural language for explaining his formal language answer, and for giving examples of pertinent erroneous answers he expected other students to make. We examined the formal language inputs with relation to the natural language ones, and we compared the automatic diagnosis done by PROBIT on the basis of the formal language, with the one we conducted manually on the base of both: formal language and the natural language explanation and examples. Examples from protocols are discussed with emphasize on the inter relationships between the two goals of an ITS’ learning environment: providing the student with useful tools to learn and providing the system with useful information for student modeling. We discuss the relevance of the research on producing cooperative explanations, to the incorporation of students’ explanations.