Diverse and Effective Synthetic Data Generation for Adaptable Zero-Shot Dialogue State Tracking

TL;DR

This paper introduces a fully automatic synthetic data generation method that significantly enhances zero-shot dialogue state tracking by creating diverse datasets across over 1,000 domains, leading to substantial performance improvements.

Contribution

The work presents a novel synthetic data generation approach that produces extensive, diverse datasets with dialogue annotations, enabling effective zero-shot DST training across many domains.

Findings

Improved Joint Goal Accuracy by 6.7% on MultiWOZ

Synthetic data enables training of smaller models with competitive performance

Generated datasets cover over 1,000 domains

Abstract

We demonstrate substantial performance gains in zero-shot dialogue state tracking (DST) by enhancing training data diversity through synthetic data generation. Existing DST datasets are severely limited in the number of application domains and slot types they cover due to the high costs of data collection, restricting their adaptability to new domains. This work addresses this challenge with a novel, fully automatic data generation approach that creates synthetic zero-shot DST datasets. Distinguished from previous methods, our approach can generate dialogues across a massive range of application domains, complete with silver-standard dialogue state annotations and slot descriptions. This technique is used to create the D0T dataset for training zero-shot DST models, encompassing an unprecedented 1,000+ domains. Experiments on the MultiWOZ benchmark show that training models on diverse synthetic data improves Joint Goal Accuracy by 6.7%, achieving results competitive with models 13.5 times larger than ours.