Global-Locally Self-Attentive Dialogue State Tracker

TL;DR

The paper introduces GLAD, a novel dialogue state tracker that combines global and local self-attention modules to improve accuracy, especially for rare states, achieving state-of-the-art results on benchmark datasets.

Contribution

GLAD is the first model to integrate global and local self-attention modules for dialogue state tracking, enhancing performance on multiple benchmarks.

Findings

Achieves 88.1% joint goal accuracy on WoZ, outperforming previous methods.

Attains 74.5% joint goal accuracy on DSTC2, surpassing prior state-of-the-art.

Significantly improves tracking of rare dialogue states.

Abstract

Dialogue state tracking, which estimates user goals and requests given the dialogue context, is an essential part of task-oriented dialogue systems. In this paper, we propose the Global-Locally Self-Attentive Dialogue State Tracker (GLAD), which learns representations of the user utterance and previous system actions with global-local modules. Our model uses global modules to share parameters between estimators for different types (called slots) of dialogue states, and uses local modules to learn slot-specific features. We show that this significantly improves tracking of rare states and achieves state-of-the-art performance on the WoZ and DSTC2 state tracking tasks. GLAD obtains 88.1% joint goal accuracy and 97.1% request accuracy on WoZ, outperforming prior work by 3.7% and 5.5%. On DSTC2, our model obtains 74.5% joint goal accuracy and 97.5% request accuracy, outperforming prior work by 1.1% and 1.0%.