计算机科学

计算机科学 ›› 2023, Vol. 50 ›› Issue (11): 185-191.doi: 10.11896/jsjkx.221000078

• 人工智能 • 上一篇    下一篇

基于核心句的端到端事件共指消解

环志刚1,2, 蒋国权1, 张玉健2, 刘浏1,3, 丁鲲1   

  1. 1 国防科技大学第六十三研究所 南京 210007
    2 东南大学网络空间安全学院 南京 211189
    3 宿迁学院信息工程学院 江苏 宿迁 223800
  • 收稿日期:2022-10-10 修回日期:2023-03-23 出版日期:2023-11-15 发布日期:2023-11-06
  • 通讯作者: 蒋国权(jianggq2001@163.com)
  • 作者简介:(zhiganghuan@seu.edu.cn)
  • 基金资助:
    中国博士后科学基金面上资助(2021MD703983);国防科技大学校科研计划项目(ZK20-46)

End-to-End Event Coreference Resolution Based on Core Sentence

HUAN Zhigang1,2, JIANG Guoquan1, ZHANG Yujian2, LIU Liu1,3, DING Kun1   

  1. 1 The Sixty-third Research Institute,National University of Defense Technology,Nanjing 210007,China
    2 School of Cyber Science and Engineering,Southeast University,Nanjing 211189,China
    3 School of Information Engineering,Suqian University,Suqian,Jiangsu 223800,China
  • Received:2022-10-10 Revised:2023-03-23 Online:2023-11-15 Published:2023-11-06
  • About author:HUAN Zhigang,born in 1997,postgra-duate,is a student member of China Computer Federation.His main research interests include knowledge graph and natural language processing.JIANG Guoquan,born in 1978,associate research fellow,master,is a member of China Computer Federation.His main research interests include equipment data engineering and knowledge graph.
  • Supported by:
    General Support from China Postdoctoral Science Foundation(2021MD703983) and Scientific Research Program of National University of Defense Technology(ZK20-46).

PDF (PC)

1928

摘要： 大多数先前的事件共指消解模型都属于成对相似度模型,通过编码两个事件提及的表示并计算相似度来判断是否共指。但是,当两个事件提及在文档内出现的位置接近时,编码其中一个事件提及的上下文表示会引入另一事件的信息,从而降低模型的性能。针对此问题,提出了一种基于核心句的端到端事件共指消解模型(End-to-end Event Coreference Resolution Based on Core Sentence,ECR-CS),该模型自动抽取事件信息并按照预先设置好的模板为每个事件提及构造核心句,利用核心句的表示代替事件提及的表示。由于核心句中只包含单个事件的信息,因此所提模型可以在编码事件表示时消除其他事件信息的干扰。此外,受到事件信息抽取工具的性能限制,构造的核心句可能会丢失事件的部分重要信息,提出利用事件在文档中的上下文表示来进行出弥补。所提模型引入了一种门控机制,将上下文嵌入向量分解为分别与核心句嵌入向量平行和正交的两个分量,平行分量可以认为是与核心句信息维度相同的信息,正交分量则是核心句中不包含的新信息。通过上下文信息和核心句信息的相关度,控制正交分量中被用来补充核心句中缺失的重要信息的新信息的量。在ACE2005数据集上进行实验,结果表明,相比最先进的模型,ECR-CS的CoNLL和AVG分数分别提升了1.76和1.04。

关键词: 事件共指消解, 门控机制, 神经网络, 预训练语言模型, 事件核心句

Abstract: Most previous event coreference resolution models belong to pairwise similarity models,which judge whether the two events are coreferences by calculating the similarity between them.However,when two event mentions appear close to each other in the document,encoding one event contextual representation will introduce information from the other event,which degrades the performance of the model.To solve the problem,an end-to-end event coreference resolution method based on core sentence(ECR-CS) is proposed.The model automatically extracts event information and constructs a core sentence for each event mention according to the preset template,and uses the core sentence representation instead of the event representation.Since the core sentence contains only the information of a single event,the model can eliminate the interference of other event information when encoding the event representation.In addition,limited by the performance of event extraction,the core sentence may lose some important information of the event.The contextual representation of the event in the document is used to make up for this problem.To supplement the missing important information in the core sentence with the contextual information,a gated mechanism is introduced to filter the noise in the contextual representation.Experiments on dataset ACE2005 show that the CoNLL and AVG scores of ECR-CS improves by 1.76 and 1.04,respectively,compared with the state-of-the-art baseline model.

Key words: Event coreference resolution, Gated mechanism, Neural network, Pre-trained language models, Event core sentence

中图分类号: 

  • TPA391
[1]LU J,NG V.Event coreference resolution:a survey of two de-cades of research[C]//Proceedings of the Twenty-Seventh International Joint Conference on Artificial Intelligence.Amsterdam:Elsevier,2018:5479-5486.
[2]MOHAMMED M,MARWA A.Efficient topic detection system for online arabic news [J].International Journal of Computer Applications,2018,180(12):7-12.
[3]CHENG L,GAO H K,WANG H B.A news event extractionmethod in Chinese and Thai languages based on dependency tree elements combined with rules [J].Software Guide,2018,17(7):49-56,63.
[4]WU M Z,MOOSAVI M S,ROTH D,et al.Coreference reaso-ning in machine reading comprehension[C]//Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing.Stroudsburg:ACL,2021:5768-5781.
[5]DEVLIN J,CHANG M W,LEE K,et al.BERT:pre-training of deep bidirectional transformers for language understanding[C]//Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics:Human Language Technologies.Stroudsburg,PA:ACL,2019:4171-4186.
[6]LU J,NG V.Span-based event coreference resolution[C]//Proceedings of the Thirty-Fifth AAAI Conference on Artificial Intelligence.Menlo Park,CA:AAAI,2021:13489-13497.
[7]LAI T M,JI H,BUI T,et al.A context-dependent gated module for incorporating symbolic semantics into event coreference re-solution[C]//Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics:Human Language Technologies.Stroudsburg,PA:ACL,2021:3491-3499.
[8]YANG B S,CAIDIE C,FRAZIER P I.A hierarchical distance-dependent bayesian model for event coreference resolution [J].Transactions of the Association for Computational Linguistics,2015(3):517-528.
[9]LIU Z Z,MITAMURA T,HOVY E H.Graph based decoding for event sequencing and coreference resolution[C]//Procee-dings of the 27th International Conference on Computational Linguistic.Stroudsburg:ACL,2018:3645-3657.
[10]CHEN Z,JI H.Graph-based event coreference resolution[C]//Proceedings of the 2009 Workshop on Graph-based Methods for Natural Language Processing.Stroudsburg:ACL,2009:54-57.
[11]KRIMAN S,JI H.Joint detection and coreference resolution of entities and events with document-level context aggregation[C]//Proceedings of the ACL-IJCNLP 2021 Student Research Workshop.Stroudsburg:ACL,2021:174-179.
[12]LU J,NG V.Constrained multi-task learning for event corefe-rence resolution[C]//Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics:Human Language Technologies.Stroudsburg,PA:ACL,2021:4504-4514.
[13]PENG H R,SONG Y Q,ROTH D.Event Detection and Core-ference with Minimal Supervision[C]//In Proceedings of the 2016 Conference on Empirical Methods in Natural Language Processing,Austin,Texas.Association for Computational Linguistics.2016:392-402.
[14]LIN Y,JI H,HUANG F.A joint neural model for information extraction with global features[C]//Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics.Stroudsburg:ACL,2020:7999-8009.
[15]LEE K,HE L H,LEWIS M,et al.End-to-end neural coreference resolution[C]//Proc of the 2017 Conference on Empirical Me-thods in Natural Language Processing.Stroudsburg:ACL,2017:188-197.
[16]CHEN Y B,XU L H,LIU K,et al.2015.Event extraction via dynamic multi-pooling convolutional neural networks[C]//Proceedings of the 53rd Annual Meeting of the Association for Computational Linguistics and the 7th International Joint Conference on Natural Language Processing,Beijing,China.Asso-ciation for Computational Linguistics,2015(1):167-176.
[17]JOSHI M,CHEN D Q,LIU Y H,et al.Spanbert:Improvingpre-training by representing and predicting spans [J].Transactions of the Association for Computational Linguistics,2020(8):64-77.
[18]VILAIN M B,BURGER J D,ABERDEEN J S,et al.A model-theoretic coreference scoring scheme[C]//Proceedings of the 6th Conference on Message Understanding.Stroudsburg:ACL,1995:45-52.
[19]BAGGA A,BALDWIN B.Algorithms for scoring coreferencechains[C]//The First Conference Language Resources and Evaluation Coreference.ELRA,1998:563-566.
[20]LUO X Q.On coreference resolution performance metrics[C]//Human Language Technology Conference and Conference on Empirical Methods in Natural Language Processing.Stroudsburg:ACL,2005:25-32.
[21]RECASENS M,HOVY E H.BLANC:implementing the randindex for coreference evaluation [J].Natural Language Engineering,2011,17(4):485-510.
Viewed
Full text


Abstract

Cited
  Shared   
  Discussed   
No Suggested Reading articles found!
渝ICP备16013121号-4
地址：重庆市渝北区洪湖西路18号    邮编：401121  
电话：023-63500828（编辑部） 023-67039612（会议/专题） 023-67039623（财务/发票）
E-mail：jsjkx12@163.com
本系统由北京玛格泰克科技发展有限公司设计开发