A Discriminative Classification-Based Approach to Information State Updates for a Multi-Domain Dialog System

Dilek Hakkani-Tur, Gokhan Tur, Larry Heck, Ashley Fidler, and Asli Celikyilmaz

Abstract

We propose a discriminative classification approach for updating the current information state of a multi-domain dialog system based on user responses. Our method uses a set of lexical and domain independent features to compare the spoken language understanding (SLU) output for the current user turn with the previous information state. We then update the information state accordingly, employing a discriminative machine learning approach. Using a data set collected from our conversational interaction system, we investigate the impact of features based on context dependent and context independent SLU tagging schemas. We show that the proposed approach outperforms two non-trivial baselines, one based on manually crafted rules and the other on classification with lexical features alone. Furthermore, such an approach allows the addition of new domains to the dialog manager in a seamless way.

Details

Publication typeInproceedings
PublisherAnnual Conference of the International Speech Communication Association (Interspeech)
> Publications > A Discriminative Classification-Based Approach to Information State Updates for a Multi-Domain Dialog System