SUMBT: Slot-Utterance Matching for Universal and Scalable Belief Tracking

TL;DR

This paper introduces SUMBT, a universal belief tracking model for dialog systems that uses attention mechanisms to relate domain-slot-types and slot-values, improving flexibility and achieving state-of-the-art accuracy.

Contribution

The paper presents a novel slot-utterance matching belief tracker that is scalable, flexible, and non-parametric, outperforming previous slot-dependent models.

Findings

Achieved state-of-the-art joint accuracy on WOZ 2.0 and MultiWOZ datasets.

Demonstrated improved performance over slot-dependent belief trackers.

Showed enhanced flexibility in handling new slot-values without retraining.

Abstract

In goal-oriented dialog systems, belief trackers estimate the probability distribution of slot-values at every dialog turn. Previous neural approaches have modeled domain- and slot-dependent belief trackers, and have difficulty in adding new slot-values, resulting in lack of flexibility of domain ontology configurations. In this paper, we propose a new approach to universal and scalable belief tracker, called slot-utterance matching belief tracker (SUMBT). The model learns the relations between domain-slot-types and slot-values appearing in utterances through attention mechanisms based on contextual semantic vectors. Furthermore, the model predicts slot-value labels in a non-parametric way. From our experiments on two dialog corpora, WOZ 2.0 and MultiWOZ, the proposed model showed performance improvement in comparison with slot-dependent methods and achieved the state-of-the-art joint accuracy.