STAR: A Schema-Guided Dialog Dataset for Transfer Learning

TL;DR

STAR is a large, schema-guided dialog dataset designed to enhance transfer learning in task-oriented dialogue systems, enabling models to generalize across tasks and domains, especially in zero-shot scenarios.

Contribution

The paper introduces the STAR dataset and novel schema-guided dialog models that improve transfer learning and zero-shot generalization in task-oriented dialogue systems.

Findings

Effective zero-shot generalization across tasks and domains

Scalable crowd-sourcing paradigm for dataset collection

Schema-guided models outperform baselines in transfer learning

Abstract

We present STAR, a schema-guided task-oriented dialog dataset consisting of 127,833 utterances and knowledge base queries across 5,820 task-oriented dialogs in 13 domains that is especially designed to facilitate task and domain transfer learning in task-oriented dialog. Furthermore, we propose a scalable crowd-sourcing paradigm to collect arbitrarily large datasets of the same quality as STAR. Moreover, we introduce novel schema-guided dialog models that use an explicit description of the task(s) to generalize from known to unknown tasks. We demonstrate the effectiveness of these models, particularly for zero-shot generalization across tasks and domains.