Abstract
We introduce inverse reinforcement learning (IRL) as an effective paradigm for training abstractive summarization models, imitating human summarization behaviors. Our IRL model estimates the reward function using a suite of important sub-rewards for summarization and concurrently optimizes the policy network. Experimental results across datasets in different domains (CNN/DailyMail and WikiHow) and various model sizes (BART-base and BART-large) demonstrate the superiority of our proposed IRL model for summarization over MLE and RL baselines. The resulting summaries exhibit greater similarity to human-crafted gold references, outperforming MLE and RL baselines on metrics such as ROUGE, coverage, novelty, compression ratio, factuality, and human evaluations.- Anthology ID:
- 2023.findings-emnlp.436
- Volume:
- Findings of the Association for Computational Linguistics: EMNLP 2023
- Month:
- December
- Year:
- 2023
- Address:
- Singapore
- Editors:
- Houda Bouamor, Juan Pino, Kalika Bali
- Venue:
- Findings
- SIG:
- Publisher:
- Association for Computational Linguistics
- Note:
- Pages:
- 6559–6570
- Language:
- URL:
- https://aclanthology.org/2023.findings-emnlp.436
- DOI:
- 10.18653/v1/2023.findings-emnlp.436
- Cite (ACL):
- Yu Fu, Deyi Xiong, and Yue Dong. 2023. Inverse Reinforcement Learning for Text Summarization. In Findings of the Association for Computational Linguistics: EMNLP 2023, pages 6559–6570, Singapore. Association for Computational Linguistics.
- Cite (Informal):
- Inverse Reinforcement Learning for Text Summarization (Fu et al., Findings 2023)
- PDF:
- https://preview.aclanthology.org/emnlp-22-attachments/2023.findings-emnlp.436.pdf