Token-level Dynamic Self-Attention Network for Multi-Passage Reading Comprehension

Yimeng Zhuang, Huadong Wang


Abstract
Multi-passage reading comprehension requires the ability to combine cross-passage information and reason over multiple passages to infer the answer. In this paper, we introduce the Dynamic Self-attention Network (DynSAN) for multi-passage reading comprehension task, which processes cross-passage information at token-level and meanwhile avoids substantial computational costs. The core module of the dynamic self-attention is a proposed gated token selection mechanism, which dynamically selects important tokens from a sequence. These chosen tokens will attend to each other via a self-attention mechanism to model long-range dependencies. Besides, convolutional layers are combined with the dynamic self-attention to enhance the model’s capacity of extracting local semantic. The experimental results show that the proposed DynSAN achieves new state-of-the-art performance on the SearchQA, Quasar-T and WikiHop datasets. Further ablation study also validates the effectiveness of our model components.
Anthology ID:
P19-1218
Volume:
Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics
Month:
July
Year:
2019
Address:
Florence, Italy
Editors:
Anna Korhonen, David Traum, Lluís Màrquez
Venue:
ACL
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
2252–2262
Language:
URL:
https://aclanthology.org/P19-1218
DOI:
10.18653/v1/P19-1218
Bibkey:
Cite (ACL):
Yimeng Zhuang and Huadong Wang. 2019. Token-level Dynamic Self-Attention Network for Multi-Passage Reading Comprehension. In Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics, pages 2252–2262, Florence, Italy. Association for Computational Linguistics.
Cite (Informal):
Token-level Dynamic Self-Attention Network for Multi-Passage Reading Comprehension (Zhuang & Wang, ACL 2019)
Copy Citation:
PDF:
https://aclanthology.org/P19-1218.pdf
Data
QUASARQUASAR-TSQuADSearchQAWikiHop