Correcting Diverse Factual Errors in Abstractive Summarization via Post-Editing and Language Model Infilling
Vidhisha Balachandran, Hannaneh Hajishirzi, William Cohen, Yulia Tsvetkov
Abstract
Abstractive summarization models often generate inconsistent summaries containing factual errors or hallucinated content. Recent works focus on correcting factual errors in generated summaries via post-editing. Such correction models are trained using adversarial non-factual summaries constructed using heuristic rules for injecting errors. However, generating non-factual summaries using heuristics often does not generalize well to actual model errors. In this work, we propose to generate hard, representative synthetic examples of non-factual summaries through infilling language models. With this data, we train a more robust fact-correction model to post-edit the summaries to improve factual consistency. Through quantitative and qualitative experiments on two popular summarization datasets— CNN/DM and XSum—we show that our approach vastly outperforms prior methods in correcting erroneous summaries. Our model—FactEdit—improves factuality scores by over ~11 points on CNN/DM and over ~31 points on XSum on average across multiple summarization models, producing more factual summaries while maintaining competitive summarization quality.- Anthology ID:
- 2022.emnlp-main.667
- Volume:
- Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing
- Month:
- December
- Year:
- 2022
- Address:
- Abu Dhabi, United Arab Emirates
- Editors:
- Yoav Goldberg, Zornitsa Kozareva, Yue Zhang
- Venue:
- EMNLP
- SIG:
- Publisher:
- Association for Computational Linguistics
- Note:
- Pages:
- 9818–9830
- Language:
- URL:
- https://aclanthology.org/2022.emnlp-main.667
- DOI:
- 10.18653/v1/2022.emnlp-main.667
- Cite (ACL):
- Vidhisha Balachandran, Hannaneh Hajishirzi, William Cohen, and Yulia Tsvetkov. 2022. Correcting Diverse Factual Errors in Abstractive Summarization via Post-Editing and Language Model Infilling. In Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing, pages 9818–9830, Abu Dhabi, United Arab Emirates. Association for Computational Linguistics.
- Cite (Informal):
- Correcting Diverse Factual Errors in Abstractive Summarization via Post-Editing and Language Model Infilling (Balachandran et al., EMNLP 2022)
- PDF:
- https://preview.aclanthology.org/landing_page/2022.emnlp-main.667.pdf