N-Shot Learning for Augmenting Task-Oriented Dialogue State Tracking
This work addresses the challenge of data scarcity in task-oriented dialogue systems, particularly for n-shot learning scenarios, though it is incremental as it builds on existing augmentation methods by leveraging structured annotations.
The paper tackled the problem of augmenting task-oriented dialogues for dialogue state tracking (DST) by introducing a framework that uses belief state annotations to create synthetic dialogues from as few as five examples. The result was significant improvements in adapting DST models to new domains and language models to the DST task, with better performance on seen values and robustness to unseen values.
Augmentation of task-oriented dialogues has followed standard methods used for plain-text such as back-translation, word-level manipulation, and paraphrasing despite its richly annotated structure. In this work, we introduce an augmentation framework that utilizes belief state annotations to match turns from various dialogues and form new synthetic dialogues in a bottom-up manner. Unlike other augmentation strategies, it operates with as few as five examples. Our augmentation strategy yields significant improvements when both adapting a DST model to a new domain, and when adapting a language model to the DST task, on evaluations with TRADE and TOD-BERT models. Further analysis shows that our model performs better on seen values during training, and it is also more robust to unseen values. We conclude that exploiting belief state annotations enhances dialogue augmentation and results in improved models in n-shot training scenarios.