STN4DST: A Scalable Dialogue State Tracking based on Slot Tagging Navigation

TL;DR

This paper introduces STN4DST, a scalable dialogue state tracking method that jointly learns slot tagging and value position prediction, improving accuracy and efficiency especially for unknown slot values.

Contribution

The paper proposes an end-to-end slot tagging navigation approach that overcomes limitations of previous DST methods by jointly learning to locate and extract slot values in dialogue contexts.

Findings

Outperforms state-of-the-art baselines on benchmark datasets

Effectively handles unknown slot values with high accuracy

Reduces error propagation compared to previous methods

Abstract

Scalability for handling unknown slot values is a important problem in dialogue state tracking (DST). As far as we know, previous scalable DST approaches generally rely on either the candidate generation from slot tagging output or the span extraction in dialogue context. However, the candidate generation based DST often suffers from error propagation due to its pipelined two-stage process; meanwhile span extraction based DST has the risk of generating invalid spans in the lack of semantic constraints between start and end position pointers. To tackle the above drawbacks, in this paper, we propose a novel scalable dialogue state tracking method based on slot tagging navigation, which implements an end-to-end single-step pointer to locate and extract slot value quickly and accurately by the joint learning of slot tagging and slot value position prediction in the dialogue context, especially for unknown slot values. Extensive experiments over several benchmark datasets show that the proposed model performs better than state-of-the-art baselines greatly.