Unsupervised Classification of Student Dialogue Acts with Query-Likelihood Clustering

Dialogue acts model the intent underlying dialogue moves. In natural language tutorial dialogue, student dialogue moves hold important information about knowledge and goals, and are therefore an integral part of providing adaptive tutoring. Automatically classifying these dialogue acts is a challenging task, traditionally addressed with supervised classification techniques requiring substantial manual time and effort. There is growing interest in unsupervised dialogue act classification to address this limitation. This paper presents a novel unsupervised framework, query-likelihood clustering, for classifying student dialogue acts. This framework combines automated natural language processing with clustering and a novel adaptation of an information retrieval technique. Evaluation against manually labeled dialogue acts on a tutorial dialogue corpus in the domain of introductory computer science demonstrates that the proposed technique outperforms existing approaches. The results indicate that this technique holds promise for automatically understanding corpora of tutorial dialogue and for building adaptive dialogue systems.