Discourse-aware semantic self-attention for narrative reading comprehension

In this work, we propose to use linguistic annotations as a basis for a Discourse-Aware Semantic Self-Attention encoder that we employ for reading comprehension on narrative texts. We extract relations between discourse units, events, and their arguments as well as coreferring mentions, using availa...

Full description

Saved in:
Bibliographic Details
Main Authors: Mihaylov, Todor (Author) , Frank, Anette (Author)
Format: Article (Journal) Chapter/Article
Language:English
Published: November 2019
In: 2019 Conference on Empirical Methods in Natural Language Processing and 9th International Joint Conference on Natural Language Processing - proceedings of the conference
Year: 2019, Pages: 2541-2552
DOI:10.18653/v1/D19-1257
Online Access:Verlag, kostenfrei, Volltext: https://doi.org/10.18653/v1/D19-1257
Verlag, kostenfrei, Volltext: https://aclanthology.org/D19-1257
Get full text
Author Notes:Todor Mihaylov, Anette Frank
Description
Summary:In this work, we propose to use linguistic annotations as a basis for a Discourse-Aware Semantic Self-Attention encoder that we employ for reading comprehension on narrative texts. We extract relations between discourse units, events, and their arguments as well as coreferring mentions, using available annotation tools. Our empirical evaluation shows that the investigated structures improve the overall performance (up to +3.4 Rouge-L), especially intra-sentential and cross-sentential discourse relations, sentence-internal semantic role relations, and long-distance coreference relations. We show that dedicating self-attention heads to intra-sentential relations and relations connecting neighboring sentences is beneficial for finding answers to questions in longer contexts. Our findings encourage the use of discourse-semantic annotations to enhance the generalization capacity of self-attention models for reading comprehension.
Item Description:Gesehen am 15.07.2024
Physical Description:Online Resource
ISBN:9781950737901
DOI:10.18653/v1/D19-1257