计算机科学

计算机科学 ›› 2026, Vol. 53 ›› Issue (6A): 250400136-10.doi: 10.11896/jsjkx.250400136

• 人工智能 • 上一篇    下一篇

基于像素差分卷积网络和注意力机制的三元组抽取

冯广1, 林健忠1, 钟婷1, 周垣桦1, 郑润庭2, 刘天翔2   

  1. 1 广东工业大学自动化学院 广州 510006
    2 广东工业大学计算机学院 广州 510006
  • 出版日期:2026-06-16 发布日期:2026-06-12
  • 通讯作者: 冯广(von@gdut.edu.cn)
  • 基金资助:
    国家自然科学基金重点项目(62237001);广东省哲学社会科学青年项目(GD23YJY08)

Triple Extraction Based on Pixel Difference Convolutional Network and Attention Mechanism

FENG Guang1, LIN Jianzhong1, ZHONG Ting1, ZHOU Yuanhua1, ZHENG Runting2, LIU Tianxiang2   

  1. 1 School of Automation,Guangdong University of Technology,Guangzhou 510006,China
    2 School of Computer Science,Guangdong University of Technology,Guangzhou 510006,China
  • Published:2026-06-16 Online:2026-06-12
  • About author:FENG Guang,born in 1973,Ph.D,professor-level senior experimenter,and master's supervisor.His main research interests include classroom streaming,big data and artificial intelligence.
  • Supported by:
    Key Program of the National Natural Science Foundation of China(62237001) and Youth Project of Guangdong Provincial Philosophy and Social Sciences(GD23YJY08).

PDF (PC)

40

摘要： 从非结构化文本中抽取关系三元组是构建知识图谱的核心环节,传统管道式和联合式抽取模型因为上下文信息捕获不足,导致语义缺失与实体边界模糊,进而引发关系冗杂和重叠问题。针对该问题,文中提出了基于像素差分卷积网络和注意力机制的关系抽取模型,该模型使用BERT对句子表示进行编码,生成对应的主体、客体、关系的句子标记表示,采用像素差分卷积网络和注意力机制,从局部和全局两个方向捕获主客体句子标记表示之间的上下文语义信息,提高实体对之间的交互性,以解决关系冗杂和重叠问题,通过双向抽取的方式减少了误差传递问题,并且加入条件归一化动态地调整句子特征,增强句子和实体间的联系,将抽取的实体对通过双仿机制进行关系预测。在纽约时报(NYT)和网络自然文本生成(WebNLG)数据集上评估了所提模型,实验结果表明,与其他基线相比,所提模型可以更好地提取重叠三元组。

关键词: 像素差分卷积, 卷积注意力机制, 双仿机制, 重叠三元组, 条件归一化

Abstract: Extracting relational triples from unstructured text is crucial for building knowledge graphs.Traditional models often suffer from relational redundancy and overlap due to insufficient contextual information capture.To tackle this,this paper proposes a relation extraction model based on pixel difference convolutional networks and attention mechanisms.It uses BERT to encode sentence representations and generate subject,object,and relation markers.By capturing contextual semantic information from local and global perspectives,the proposed model enhances entity pair interaction,reduces error propagation via bidirectional extraction,and strengthens sentence-entity connections through conditional normalization.A double imitation mechanism is employed to predict triples.Experiments on NYT and WebNLG datasets show the proposed model outperforms baselines in extracting overlapping triples.

Key words: Pixel difference convolution, Convolutional block attention module, Biaffine attention mechanism, Overlapping triples, Conditional normalization

中图分类号: 

  • TP391.1
[1] WU H,LI S,GAO Y,et al.Natural language processing in educational research:The evolution of research topics[J].Education and Information Technologies,2024,29(17):23271-23297.
[2] JI S,PAN S,CAMBRIA E,et al.A survey on knowledgegraphs:Representation,acquisition,and applications[J].IEEE Transactions on Neural Networks and Learning Systems,2021,33(2):494-514.
[3] MARTINEZ-RODRIGUEZ J L,HOGAN A,LOPEZ-AREVALO I.Information extraction meets the semantic web:a survey[J].Semantic Web,2020,11(2):255-335.
[4] E H H,ZHANG W J,XIAO S Q,et al.Review of entity relation Extraction in Deep Learning[J].Journal of Software,2019,30(6):1793-1818.
[5] ZENG X,ZENG D,HE S,et al.Extracting Relational Facts by an End-to-End Neural Model with Copy Mechanism[C]//Proceedings of the 56th Annual Meeting of the Association for Computational Linguistics(Volume 1:Long Papers).2018.DOI:10.18653/v1/P18-1047.
[6] GIANNIS B,JOHANNES D,THOMAS D,et al.Joint entity recognition and relation extraction as a multi-head selection problem[J].Expert Systems with Application,2018,114(DEC.):34-45.
[7] PENG B,LI Y D,GONG X F,et al.An entity relation extraction method based on heterogeneous graph neural network and text semantic enhancement[J].Computer Science,2024,51(6A):230700071-5.
[8] YU B,ZHANG Z,SHU X,et al.Joint Extraction of Entities and Relations Based on a Novel Decomposition Strategy[C]//24th European Conference on Artificial Intelligence:ECAI 2020,Including 10th Conference on Prestigious Applications of Artificial Intelligence(PAIS 2020),Santiago de Compostela,Spain, 2020.
[9] ZHANG Y S,LIU S K,LIU Y,et al.A review of entity relationship joint extraction based on deep learning[J].Acta Electronica Sinica,2023,51(4):1093-1116.
[10] YUAN Y,ZHOU X,PAN S,et al.A Relation-Specific Attention Network for Joint Entity and Relation Extraction[C]//Twenty-Ninth International Joint Conference on Artificial Intelligence and Seventeenth Pacific Rim International Conference on Artificial Intelligence.2020.
[11] MA Y,HIRAOKA T,OKAZAKI N.Joint entity and relationextraction based on table labeling using convolutional neural networks[C]//Proceedings of the sixth workshop on structured prediction for NLP.2022:11-21.
[12] WEI Z,SU J,WANG Y,et al.A Novel Cascade Binary Tagging Framework for Relational Triple Extraction[C]//Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics.2020.
[13] YE H,ZHANG N,DENG S,et al.Contrastive Triple Extraction with Generative Transformer[C]//National Conference on Artificial Intelligence.2021.
[14] LU S,ZHANG W,CHENG L.Combining Sentence-based Relational Features with Biaffine Mechanism for Triple Extraction[C]//Proceedings of the 2024 16th International Conference on Machine Learning and Computing.2024:121-127.
[15] ZENG W,XI X F,CUI Z M.Joint extraction of entities and relations in short texts based on a parallel residual dilated convolutional network[J].Modern Electronics Technique,2025(2).
[16] LI Z M,ZHANG Y,YIN F,et al.Entity relation extraction based on self-attention and piecewise convolutional neural network[J].Journal of South-Central Minzu University(Natural Science Edition),2022(3).
[17] CHEN Y,WANG K,YANG W,et al.A Multi-Channel Deep Neural Network for Relation Extraction[J].IEEE Access,2020,PP(99):1-1.
[18] SU Z,LIU W,YU Z,et al.Pixel Difference Networks for Efficient Edge Detection[J].2021.DOI:10.48550/arXiv.2108.07009.
[19] WOO S,PARK J,LEE J Y,et al.CBAM:Convolutional block attention module[C]//Proceedings of the European Conference on Computer Vision.2018:3-19.
[20] TANG X L,DING H C,TANG Y Z,et al.Entity relation extraction based on heterogeneous graph and semantic fusion[J].Experimental Technology & Management,2024,41(8).
[21] TIANXT,JING L P,HE L,et al.StereoRel:Relational triple extraction from a stereoscopic perspective[C]//Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing(Volume 1:Long Papers).2021:4851-4861.
[22] WANG Y C,YU B,ZHANG Y,et al.TPLinker:Single-stagejoint extraction of entities and relations through token pair linking[C]//Proceedings of the 28th international conference on computational linguistics.2020.
[23] LI Z J,HAN R R,LI C H,et al.An entity relation extraction method integrating pre-trained models and attention mechanism[J].Computer Science and Exploration,2023,17(6):1453-1462.
Viewed
Full text


Abstract

Cited
  Shared   
  Discussed   
No Suggested Reading articles found!
渝ICP备16013121号-4
地址：重庆市两江新区洪湖西路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