Journal of Computer Applications ›› 2024, Vol. 44 ›› Issue (10): 3288-3293.DOI: 10.11772/j.issn.1001-9081.2023101542

• The 40th CCF National Database Conference (NDBC 2023) • Previous Articles     Next Articles

Machine reading comprehension event detection based on relation-enhanced graph convolutional network

Wanting JI, Wenyi LU, Yuhang MA, Linlin DING, Baoyan SONG, Haolin ZHANG()   

  1. College of Information,Liaoning University,Shenyang Liaoning 110036,China
  • Received:2023-11-10 Revised:2024-01-01 Accepted:2024-01-03 Online:2024-10-15 Published:2024-10-10
  • Contact: Haolin ZHANG
  • About author:JI Wanting, born in 1992, Ph. D., lecturer. Her research interests include natural language processing, deep learning.
    LU Wenyi, born in 1998, M. S. candidate. Her research interests include natural language processing.
    MA Yuhang, born in 1999, M. S. candidate. His research interests include natural language processing.
    DING Linlin, born in 1983, Ph. D., professor. Her research interests include big data management, distributed data management, uncertain data management.
    SONG Baoyan, born in 1965, Ph. D., professor. Her research interests include database, big data management, graph data management.
  • Supported by:
    Applied Basic Research Program of Liaoning Province(2022JH2/101300250);National Key R&D Program of China(2023YFC3304900);Doctoral Startup Project of Liaoning Natural Science Foundation(2023-BS-085);University Basic Scientific Research Project of Education Department of Liaoning Province(Engineering┫ (Leading the Charge with Open Competition Project for Local Service) ┣JYTMS20230761)

基于关系增强图卷积网络的机器阅读理解式事件检测

纪婉婷, 鲁闻一, 马宇航, 丁琳琳, 宋宝燕, 张浩林()   

  1. 辽宁大学 信息学院,沈阳 110036
  • 通讯作者: 张浩林
  • 作者简介:纪婉婷(1992—),女,辽宁沈阳人,讲师,博士,CCF会员,主要研究方向:自然语言处理、深度学习
    鲁闻一(1998—),女,辽宁铁岭人,硕士研究生,CCF会员,主要研究方向:自然语言处理
    马宇航(1999—),男,河南焦作人,硕士研究生,CCF会员,主要研究方向:自然语言处理
    丁琳琳(1983—),女,辽宁阜新人,教授,博士,CCF会员,主要研究方向:大数据管理、分布式数据管理、不确定数据管理
    宋宝燕(1965—),女,辽宁开原人,教授,博士,CCF高级会员,主要研究方向:数据库、大数据管理、图数据管理
    张浩林(1979—),男,辽宁沈阳人,博士研究生,主要研究方向:深度学习、大数据管理 haolinzhang@163.com
  • 基金资助:
    辽宁省应用基础研究计划项目(2022JH2/101300250);国家重点研发计划项目(2023YFC3304900);辽宁省自然科学基金博士启动项目(2023?BS?085);辽宁省教育厅高校基本科研项目(理工类)(揭榜挂帅服务地方项目)(JYTMS20230761)

Abstract:

Aiming at the problem that existing machine reading comprehension-based event detection models are difficult to mine the long-distance dependencies between keywords when facing long text context with complex syntactic relations, a Machine Reading Comprehension event detection model based on Relation-Enhanced Graph Convolutional Network (MRC-REGCN) was proposed. Firstly, a pre-trained language model was utilized to jointly encode the question and the text to obtain word vector representations incorporating the priory information. Secondly, the dynamic relationship was introduced to enhance the label information, and an REGCN was utilized to deeply learn syntactic dependencies between words and enhance the model’s ability to perceive the syntactic structure of long text. Finally, the probability distributions of the textual words under all event types were obtained by the multi-classifier. Experimental results on the ACE2005 English corpus show that the F1 score of the proposed model on trigger classification is improved by 2.49% and 1.23% compared to the comparable machine reading comprehension-based model named EEQA (Event Extraction by Answering (almost) natural Questions) and the best baseline model named DEGREE (Data-Efficient GeneRation-based Event Extraction) respectively, which verify that the MRC-REGCN has a better performance in performing event detection.

Key words: machine reading comprehension, event detection, Graph Convolutional Network (GCN), syntactic dependency relation, trigger classification

摘要:

在面对具有复杂句法关系的长文本上下文时,现有机器阅读理解式事件检测模型难以挖掘关键词之间长距离依赖关系。针对上述问题,提出一种基于关系增强图卷积网络(REGCN)的机器阅读理解式事件检测模型(MRC-REGCN)。首先,利用预训练语言模型对问题和文本进行联合编码,得到融入先验信息的单词向量表示;其次,引入动态的关系增强标签信息,并利用REGCN深入学习单词之间的句法依存关系,增强模型对长文本句法结构的感知能力;最后,利用多分类器得到文本单词在所有事件类型下的概率分布。在ACE2005英文语料上的实验结果表明,所提模型在触发词分类上的F1分值相较于同类机器阅读理解模型EEQA(Event Extraction by Answering (almost) natural Questions)和最佳基线模型DEGREE(Data-Efficient GeneRation-based Event Extraction)分别提升了2.49%和1.23%,验证了MRC-REGCN具有更好的事件检测性能。

关键词: 机器阅读理解, 事件检测, 图卷积网络, 句法依存关系, 触发词分类

CLC Number: