Keyphrase Generation via Soft and Hard Semantic Corrections

Guangzhen Zhao, Guoshun Yin, Peng Yang, Yu Yao


Abstract
Keyphrase generation aims to generate a set of condensed phrases given a source document. Although maximum likelihood estimation (MLE) based keyphrase generation methods have shown impressive performance, they suffer from the bias on the source-prediction sequence pair and the bias on the prediction-target pair. To tackle the above biases, we propose a novel correction model CorrKG on top of the MLE pipeline, where the biases are corrected via the optimal transport (OT) and a frequency-based filtering-and-sorting (FreqFS) strategy. Specifically, OT is introduced as soft correction to facilitate the alignment of salient information and rectify the semantic bias in the source document and predicted keyphrases pair. An adaptive semantic mass learning scheme is conducted on the vanilla OT to achieve a proper pair-wise optimal transport procedure, which promotes the OT learning brought by rectifying semantic masses dynamically. Besides, the FreqFS strategy is designed as hard correction to reduce the bias of predicted and ground truth keyphrases, and thus to generate accurate and sufficient keyphrases. Extensive experiments over multiple benchmark datasets show that our model achieves superior keyphrase generation as compared with the state-of-the-arts.
Anthology ID:
2022.emnlp-main.529
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:
7757–7768
Language:
URL:
https://aclanthology.org/2022.emnlp-main.529
DOI:
10.18653/v1/2022.emnlp-main.529
Bibkey:
Cite (ACL):
Guangzhen Zhao, Guoshun Yin, Peng Yang, and Yu Yao. 2022. Keyphrase Generation via Soft and Hard Semantic Corrections. In Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing, pages 7757–7768, Abu Dhabi, United Arab Emirates. Association for Computational Linguistics.
Cite (Informal):
Keyphrase Generation via Soft and Hard Semantic Corrections (Zhao et al., EMNLP 2022)
Copy Citation:
PDF:
https://aclanthology.org/2022.emnlp-main.529.pdf