Abstract
Training task-oriented dialogue systems typically requires turn-level annotations for interacting with their APIs: e.g. a dialogue state and the system actions taken at each step. These annotations can be costly to produce, error-prone, and require both domain and annotation expertise. With advances in LLMs, we hypothesize that unlabeled data and a schema definition are sufficient for building a working task-oriented dialogue system, completely unsupervised. We consider a novel unsupervised setting of only (1) a well-defined API schema (2) a set of unlabeled dialogues between a user and agent. We propose an innovative approach using expectation-maximization (EM) that infers turn-level annotations as latent variables using a noisy channel model to build an end-to-end dialogue agent. Evaluating our approach on the MultiWOZ benchmark, our method more than doubles the dialogue success rate of a strong GPT-3.5 baseline.- Anthology ID:
- 2024.emnlp-main.473
- Volume:
- Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing
- Month:
- November
- Year:
- 2024
- Address:
- Miami, Florida, USA
- Editors:
- Yaser Al-Onaizan, Mohit Bansal, Yun-Nung Chen
- Venue:
- EMNLP
- SIG:
- Publisher:
- Association for Computational Linguistics
- Note:
- Pages:
- 8283–8300
- Language:
- URL:
- https://aclanthology.org/2024.emnlp-main.473
- DOI:
- 10.18653/v1/2024.emnlp-main.473
- Cite (ACL):
- Brendan King and Jeffrey Flanigan. 2024. Unsupervised End-to-End Task-Oriented Dialogue with LLMs: The Power of the Noisy Channel. In Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing, pages 8283–8300, Miami, Florida, USA. Association for Computational Linguistics.
- Cite (Informal):
- Unsupervised End-to-End Task-Oriented Dialogue with LLMs: The Power of the Noisy Channel (King & Flanigan, EMNLP 2024)
- PDF:
- https://preview.aclanthology.org/dois-2013-emnlp/2024.emnlp-main.473.pdf