Abstract
This paper introduces a novel Self-supervised Fine-grained Dialogue Evaluation framework (SelF-Eval). The core idea is to model the correlation between turn quality and the entire dialogue quality. We first propose a novel automatic data construction method that can automatically assign fine-grained scores for arbitrarily dialogue data. Then we train SelF-Eval with a multi-level contrastive learning schema which helps to distinguish different score levels. Experimental results on multiple benchmarks show that SelF-Eval is highly consistent with human evaluations and better than the state-of-the-art models. We give a detailed analysis of the experiments in this paper. Our code is available on GitHub.- Anthology ID:
- 2022.coling-1.39
- Volume:
- Proceedings of the 29th International Conference on Computational Linguistics
- Month:
- October
- Year:
- 2022
- Address:
- Gyeongju, Republic of Korea
- Venue:
- COLING
- SIG:
- Publisher:
- International Committee on Computational Linguistics
- Note:
- Pages:
- 485–495
- Language:
- URL:
- https://aclanthology.org/2022.coling-1.39
- DOI:
- Cite (ACL):
- Longxuan Ma, Ziyu Zhuang, Weinan Zhang, Mingda Li, and Ting Liu. 2022. SelF-Eval: Self-supervised Fine-grained Dialogue Evaluation. In Proceedings of the 29th International Conference on Computational Linguistics, pages 485–495, Gyeongju, Republic of Korea. International Committee on Computational Linguistics.
- Cite (Informal):
- SelF-Eval: Self-supervised Fine-grained Dialogue Evaluation (Ma et al., COLING 2022)
- PDF:
- https://preview.aclanthology.org/ingestion-script-update/2022.coling-1.39.pdf
- Code
- royny/self-eval
- Data
- ConvAI2, DailyDialog, DailyDialog++, FED