AutoTutor’s Coverage of Expectations during Tutorial Dialogue

Art Graesser, Andrew Olney, Matthew Ventura, and G. Tanner Jackson, University of Memphis

AutoTutor is a learning environment with an animated agent that tutors students by holding a conversation in natural language. AutoTutor presents challenging questions and then engages in mixed initiative dialogue that guides the student in building an answer. AutoTutor uses latent semantic analysis (LSA) as a major component that statistically represents world knowledge and tracks whether particular expectations and misconceptions are expressed by the learner. This paper describes AutoTutor, reports some analyses on the adequacy of the LSA component, and proposes some improvements in computing the coverage of particular expectations and misconceptions.


This page is copyrighted by AAAI. All rights reserved. Your use of this site constitutes acceptance of all of AAAI's terms and conditions and privacy policy.