Share on Facebook Tweet on Twitter Share on LinkedIn Share by email
Discovering Latent Structure in Task-Oriented Dialogues

Ke Zhai and Jason Williams

Abstract

A key challenge for computational conversation models is to discover latent structure in task-oriented dialogue, since it provides a basis for analysing, evaluating, and building conversational systems. We propose three new unsupervised models to discover latent structures in task-oriented dialogues. Our methods synthesize hidden Markov models (for underlying state) and topic models (to connect words to states). We apply them to two real, non-trivial datasets: human-computer spoken dialogues in bus query service, and human-human text-based chats from a live technical support service. We show that our models extract meaningful state representations and dialogue structures consistent with human annotations. Quantitatively, we show our models achieve superior performance on held-out log likelihood evaluation and an ordering task.

Details

Publication typeInproceedings
Published inProceedings of ACL 2014
PublisherAssociation for Computational Linguistics
> Publications > Discovering Latent Structure in Task-Oriented Dialogues