STAR: A Schema-Guided Dialog Dataset for Transfer Learning
This addresses the problem of limited generalization in task-oriented dialog systems for researchers and developers, though it is incremental as it builds on existing transfer learning approaches.
The authors introduced STAR, a schema-guided task-oriented dialog dataset with 127,833 utterances across 5,820 dialogs in 13 domains, designed to enable transfer learning, and proposed novel schema-guided models that achieve effective zero-shot generalization across tasks and domains.
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.