Context-Dependent Transitions in Tutoring Discourse

Beverly Park Woolf, David D. McDonald

Abstract: Successful machine tutoring, like other forms of human-machine discourse, requires sophisticated communication skills and a deep understanding of the student's knowledge. A system must have the ability to reason about a student's knowledge and to assess the effect of the discourse on him. In this paper we describe Meno-tutor, a LISP program that deliberately plans the rhetorical structure of its output and customizes its responses to the level of understanding of the individual student.