Unsupervised classification of dialogue acts using a dirichlet process mixture model

Abstract

In recent years Dialogue Acts have become a popular means of modelling the communicative intentions of human and machine utterances in many modern dialogue systems. Many of these systems rely heavily on the availability of dialogue corpora that have been annotated with Dialogue Act labels. The manual annotation of dialogue corpora is both tedious and expensive. Consequently, there is a growing interest in unsupervised systems that are capable of automating the annotation process. This paper investigates the use of a Dirichlet Process Mixture Model as a means of clustering dialogue utterances in an unsupervised manner. These clusters can then be analysed in terms of the possible Dialogue Acts that they might represent. The results presented here are from the application of the Dirichlet Process Mixture Model to the Dihana corpus.

Automatic dialogue act recognition with syntactic features

“…Hence, the authors of [25] propose a predictive paradigm where dialogue act models are first trained on a small-size corpus and used afterwards to predict future sentences or dialogue acts. In a related vein, unsupervised dialogue act tagging of unlabelled text has recently raised a lot of attention [26,27], but we will limit ourselves in the following on supervised approaches.…”

Discovering Latent Structure in Task-Oriented Dialogues

“…By contrast, unsupervised methods operate only on the observable signal (e.g. words) and are estimated without labels or their attendant limitations (Crook et al, 2009). They are particularly relevant because conversation is a temporal process where models are trained to infer a latent state which evolves as the dialogue progresses (Bangalore et al, 2006;Traum and Larsson, 2003).…”

Combining Task and Dialogue Streams in Unsupervised Dialogue Act Models

“…To overcome these drawbacks, the field has recently seen growing momentum surrounding unsupervised approaches, which do not require any manual labels during model training (Crook et al, 2009;Joty et al, 2011;Lee et al, 2013). A variety of unsupervised machine learning techniques have been investigated for dialogue act classification, and each line of investigation has explored which features best support this goal.…”

“…The task of automatic dialogue act classification has been extensively studied for decades within several domains including train fares and timetables (Allen et al, 1995;Core and Allen, 1997;Crook et al, 2009;Traum, 1999), virtual personal assistants (Chen and Di Eugenio, 2013), conversational telephone speech (Stolcke et al, 2000), Wikipedia talk pages (Ferschke et al, 2012) and as in the case of this paper, tutorial dialogue (Serafin and Di Eugenio, 2004;Forbes-Riley and Litman, 2005;Boyer et al, 2011;Dzikovska et al, 2013).…”