Hendrix: A Conversational Intelligent Tutoring System for Java Programming

Holmes, M, Latham, A, Crockett, K, Lewin, C and O'Shea, J (2015) Hendrix: A Conversational Intelligent Tutoring System for Java Programming. In: UKCI 2015 15th UK Workshop on Computational Intelligence, 07 September 2015 - 09 September 2015, University of Exeter.

Preview

Download (509kB) | Preview

Abstract

This paper proposes a novel generic architecture for a conversational intelligent tutoring system named Hendrix. Hendrix mimics a human tutor by guiding a learner through a given knowledge domain using natural language. Hendrix converses with a learner to identify gaps in knowledge through questioning, expanding the curriculum when gaps in knowledge are identified. Hendrix supports learners by detecting questions and providing definitions and examples. Hendrix novel architecture uses a graph of concepts to dynamically generate tutorials. Hendrix uses both syntactic and semantic language analysis to extract and match information from learner utterances. Hendrix’ two loop algorithm is dependent on identifying the short term goal a learner in each conversational turn. In a pilot study, Hendrix correctly classified the utterance type of 91% of input sentences, marked 94.5% of question answers correctly, and was rated 3.93 out of 5 for user satisfaction