Point or Generate Dialogue State Tracker

TL;DR

The paper introduces POGD, a novel dialogue state tracker that explicitly identifies expressed slot values and generates implicit ones, sharing parameters across slots for scalability and improved generalization.

Contribution

It proposes a dual approach combining pointing and generating for dialogue state tracking, with shared parameters and multi-task training for better scalability and generalization.

Findings

Achieves state-of-the-art results on WoZ 2.0 and MultiWoZ 2.0 datasets.

Demonstrates strong generalization to unseen values and new slots.

Effectively combines explicit pointing and implicit generation in dialogue state tracking.

Abstract

Dialogue state tracking is a key part of a task-oriented dialogue system, which estimates the user's goal at each turn of the dialogue. In this paper, we propose the Point-Or-Generate Dialogue State Tracker (POGD). POGD solves the dialogue state tracking task in two perspectives: 1) point out explicitly expressed slot values from the user's utterance, and 2) generate implicitly expressed ones based on slot-specific contexts. It also shares parameters across all slots, which achieves knowledge sharing and gains scalability to large-scale across-domain dialogues. Moreover, the training process of its submodules is formulated as a multi-task learning procedure to further promote its capability of generalization. Experiments show that POGD not only obtains state-of-the-art results on both WoZ 2.0 and MultiWoZ 2.0 datasets but also has good generalization on unseen values and new slots.