In Conclusion Not Repetition: Comprehensive Abstractive Summarization with Diversified Attention Based on Determinantal Point Processes
Lei Li, Wei Liu, Marina Litvak, Natalia Vanetik, Zuying Huang
Abstract
Various Seq2Seq learning models designed for machine translation were applied for abstractive summarization task recently. Despite these models provide high ROUGE scores, they are limited to generate comprehensive summaries with a high level of abstraction due to its degenerated attention distribution. We introduce Diverse Convolutional Seq2Seq Model(DivCNN Seq2Seq) using Determinantal Point Processes methods(Micro DPPs and Macro DPPs) to produce attention distribution considering both quality and diversity. Without breaking the end to end architecture, DivCNN Seq2Seq achieves a higher level of comprehensiveness compared to vanilla models and strong baselines. All the reproducible codes and datasets are available online.- Anthology ID:
- K19-1077
- Original:
- K19-1077v1
- Version 2:
- K19-1077v2
- Volume:
- Proceedings of the 23rd Conference on Computational Natural Language Learning (CoNLL)
- Month:
- November
- Year:
- 2019
- Address:
- Hong Kong, China
- Editors:
- Mohit Bansal, Aline Villavicencio
- Venue:
- CoNLL
- SIG:
- SIGNLL
- Publisher:
- Association for Computational Linguistics
- Note:
- Pages:
- 822–832
- Language:
- URL:
- https://aclanthology.org/K19-1077
- DOI:
- 10.18653/v1/K19-1077
- Cite (ACL):
- Lei Li, Wei Liu, Marina Litvak, Natalia Vanetik, and Zuying Huang. 2019. In Conclusion Not Repetition: Comprehensive Abstractive Summarization with Diversified Attention Based on Determinantal Point Processes. In Proceedings of the 23rd Conference on Computational Natural Language Learning (CoNLL), pages 822–832, Hong Kong, China. Association for Computational Linguistics.
- Cite (Informal):
- In Conclusion Not Repetition: Comprehensive Abstractive Summarization with Diversified Attention Based on Determinantal Point Processes (Li et al., CoNLL 2019)
- PDF:
- https://preview.aclanthology.org/corrections-2024-07/K19-1077.pdf
- Code
- thinkwee/DPP_CNN_Summarization
- Data
- BigPatent, CNN/Daily Mail, NEWSROOM, WikiHow