Modeling Instance Interactions for Joint Information Extraction with Neural High-Order Conditional Random Field
2023
会议录名称PROCEEDINGS OF THE ANNUAL MEETING OF THE ASSOCIATION FOR COMPUTATIONAL LINGUISTICS
ISSN0736-587X
卷号1
页码13695-13710
发表状态已发表
摘要

Prior works on joint Information Extraction (IE) typically model instance (e.g., event triggers, entities, roles, relations) interactions by representation enhancement, type dependencies scoring, or global decoding. We find that the previous models generally consider binary type dependency scoring of a pair of instances, and leverage local search such as beam search to approximate global solutions. To better integrate cross-instance interactions, in this work, we introduce a joint IE framework (CRFIE) that formulates joint IE as a high-order Conditional Random Field. Specifically, we design binary factors and ternary factors to directly model interactions between not only a pair of instances but also triplets. Then, these factors are utilized to jointly predict labels of all instances. To address the intractability problem of exact high-order inference, we incorporate a high-order neural decoder that is unfolded from a mean-field variational inference method, which achieves consistent learning and inference. The experimental results show that our approach achieves consistent improvements on three IE tasks compared with our baseline and prior work. © 2023 Association for Computational Linguistics.

会议录编者/会议主办者Bloomberg Engineering ; et al. ; Google Research ; LIVEPERSON ; Meta ; Microsoft
关键词Computational linguistics Image segmentation Information retrieval Random processes Beam search Binary factors Directly model Event trigger Global solutions High-order Higher-order Joint information Local search Random fields
会议名称61st Annual Meeting of the Association for Computational Linguistics, ACL 2023
会议地点Toronto, ON, Canada
会议日期July 9, 2023 - July 14, 2023
收录类别EI
语种英语
出版者Association for Computational Linguistics (ACL)
EI入藏号20234314933846
EI主题词Decoding
EI分类号721.1 Computer Theory, Includes Formal Logic, Automata Theory, Switching Theory, Programming Theory ; 723.2 Data Processing and Image Processing ; 903.3 Information Retrieval and Use ; 922.1 Probability Theory
原始文献类型Conference article (CA)
文献类型会议论文
条目标识符https://kms.shanghaitech.edu.cn/handle/2MSLDSTB/345814
专题信息科学与技术学院_博士生
信息科学与技术学院_PI研究组_屠可伟组
通讯作者Zheng, Zilong; Tu, Kewei
作者单位
1.Beijing Institute for General Artificial Intelligence (BIGAI), Beijing, China;
2.ShanghaiTech University, Shanghai, China;
3.Beijing Jiaotong University, Beijing, China
第一作者单位上海科技大学
通讯作者单位上海科技大学
推荐引用方式
GB/T 7714
Jia, Zixia,Yan, Zhaohui,Han, Wenjuan,et al. Modeling Instance Interactions for Joint Information Extraction with Neural High-Order Conditional Random Field[C]//Bloomberg Engineering, et al., Google Research, LIVEPERSON, Meta, Microsoft:Association for Computational Linguistics (ACL),2023:13695-13710.
条目包含的文件
文件名称/大小 文献类型 版本类型 开放类型 使用许可
个性服务
查看访问统计
谷歌学术
谷歌学术中相似的文章
[Jia, Zixia]的文章
[Yan, Zhaohui]的文章
[Han, Wenjuan]的文章
百度学术
百度学术中相似的文章
[Jia, Zixia]的文章
[Yan, Zhaohui]的文章
[Han, Wenjuan]的文章
必应学术
必应学术中相似的文章
[Jia, Zixia]的文章
[Yan, Zhaohui]的文章
[Han, Wenjuan]的文章
相关权益政策
暂无数据
收藏/分享
所有评论 (0)
暂无评论
 

除非特别说明,本系统中所有内容都受版权保护,并保留所有权利。