My science tutor: Learning science with a conversational virtual tutor

Abstract

This paper presents a conversational, multimedia, virtual science tutor for elementary school students. It is built using state of the art speech recognition and spoken language understanding technology. This virtual science tutor is unique in that it elicits self-explanations from students for various science phenomena by engaging them in spoken dialogs and guided by illustrations, animations and interactive simulations. There is a lot of evidence that self-explanation works well as a tutorial paradigm, Summative evaluations indicate that students are highly engaged in the tutoring sessions, and achieve learning outcomes equivalent to expert human tutors. Tutorials are developed through a process of recording and annotating data from sessions with students, and then updating tutor models. It enthusiastically supported by students and teachers. Teachers report that it is feasible to integrate into their curriculum.