Incorporating Option and Out-of-domain Knowledge for Multi-choice Machine Reading Comprehension

Yuan Xu, Shumin Shi*, Heyan Huang

*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

Multi-choice Machine Reading Comprehension (MRC) requires the model to select the correct answer from a set of answer candidates given the corresponding passage and question. Previous studies mainly focus on complex matching networks to model the relationship among options, passage and question. However, these models obtain little improvement over the powerful Pre-trained Language Models (PLMs). In this paper, we propose a simple method to incorporate option knowledge from PLMs and introduce out-of-domain knowledge by multi-task learning skillfully. Our approach obtains state-of-the-art results on Chinese multi-choice MRC dataset ReCO and also effectively improves the performance on C3.

Original languageEnglish
Title of host publicationProceedings of 2021 7th IEEE International Conference on Cloud Computing and Intelligence Systems, CCIS 2021
EditorsDeyi Li, Mengqi Zhou, Weining Wang, Yaru Zou, Meng Luo, Qian Zhang
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages493-497
Number of pages5
ISBN (Electronic)9781665441490
DOIs
Publication statusPublished - 2021
Event7th IEEE International Conference on Cloud Computing and Intelligence Systems, CCIS 2021 - Xi'an, China
Duration: 7 Nov 20218 Nov 2021

Publication series

NameProceedings of 2021 7th IEEE International Conference on Cloud Computing and Intelligence Systems, CCIS 2021

Conference

Conference7th IEEE International Conference on Cloud Computing and Intelligence Systems, CCIS 2021
Country/TerritoryChina
CityXi'an
Period7/11/218/11/21

Keywords

  • Multi-choice MRC
  • Multi-task Learning
  • Option Knowledge

Fingerprint

Dive into the research topics of 'Incorporating Option and Out-of-domain Knowledge for Multi-choice Machine Reading Comprehension'. Together they form a unique fingerprint.

Cite this