计算机科学

计算机科学 ›› 2026, Vol. 53 ›› Issue (1): 271-277.doi: 10.11896/jsjkx.241100069

• 人工智能 • 上一篇    下一篇

基于元学习的跨语言知识图谱实体对齐框架

陈壮壮1, 邓怡辰3, 余敦辉1,2, 肖奎1,2   

  1. 1 湖北大学计算机与信息工程学院 武汉 430062;
    2 大数据智能分析与行业应用湖北省重点实验室 武汉 430062;
    3 武昌首义学院 武汉 430064
  • 收稿日期:2024-11-12 修回日期:2025-02-11 发布日期:2026-01-08
  • 通讯作者: 邓怡辰(104238327@qq.com)
  • 作者简介:(654414209@qq.com)
  • 基金资助:
    国家自然科学基金(62377009)

Cross-language Knowledge Graph Entity Alignment Based on Meta-learning

CHEN Zhuangzhuang1, DENG Yichen3, YU Dunhui1,2, XIAO Kui1,2   

  1. 1 School of Computer Science and Information Engineering, Hubei University, Wuhan 430062, China;
    2 Hubei Key Laboratory of Big Data Intelligent Analysis and Application, Wuhan 430062, China;
    3 Wuchang Shouyi College, Wuhan 430064, China
  • Received:2024-11-12 Revised:2025-02-11 Online:2026-01-08
  • About author:CHEN Zhuangzhuang,born in 1999,postgraduate.His main research interest is knowledge graph.
    DENG Yichen,born in 1993,master,lecturer.Her main research interest is embedded development.
  • Supported by:
    National Natural Science Foundation of China(62377009).

PDF (PC)

11

摘要： 跨语言知识图谱实体对齐是连接不同语言知识图谱的关键步骤,在多语言信息检索、数据融合等任务中有重要作用。然而,现有的实体对齐方法依赖知识图谱中的多种信息,难以很好地处理稀疏知识图谱实体对齐任务,并且对新的语言的适应性较差。针对该问题,提出了基于元学习的跨语言实体对齐框架。该框架总体分为外循环与内循环两个阶段:在外循环阶段,通过基于任务相似度的采样方法选取出多个任务,然后对模型进行多任务联合训练,构建教师模型;在内循环阶段,利用外循环阶段训练好的教师模型指导学生模型进行训练和实体对齐任务,提升学生模型实体对齐的性能和泛化性。在SRPRS和WK31-60K数据集上的实验结果表明,所提框架在实体对齐问题中,Hits@1指标平均提升3.5%,Hits@10指标平均提升4.0%,MRR指标平均提升6.3%。

关键词: 元学习, 跨语言知识图谱, 实体对齐, 外循环, 内循环, 泛化能力

Abstract: Cross-language knowledge graph entity alignment is a key step in connecting knowledge graphs of different languages,and it plays an important role in tasks such as multilingual information retrieval and data fusion.However,the existing entity alignment methods rely on a variety of information in the knowledge graph,which cannot handle the entity alignment task of the sparse knowledge graph well,and has poor adaptability to new languages.To solve this problem,a cross-language entity alignment framework based on meta-learning is proposed.The framework is generally divided into two stages,the outer loop and the inner loop.In the outer loop stage,multiple tasks are selected through the sampling method based on task similarity,and then the model is jointly trained with multiple tasks to construct the teacher model.In the inner loop stage,the teacher model trained in the outer loop stage is used to guide the student model to carry out the training and entity alignment tasks,in order to improve the entity alignment performance and generalization of the student model.Experimental results on the SRPRS and WK31-60K dataset show that the proposed framework improves the Hits@1 index by 3.5%,Hits@10 index by 4.0%,and MRR index by 6.3% on average in the entity alignment problem.

Key words: Meta-learning, Cross-language knowledge graph, Entity alignment, Outer loop, Inner loop, Generalization ability

中图分类号: 

  • TP391
[1]ZHANG F,YANG L Y,LI J W,et al.A Review of EntityAlignment Research[J].Journal of Computer Science,2022,45(6):1195-1225.
[2]CAO Y X,WANG X,HE X G,et al.Unifying knowledge graph learning and recommendation:Towards a better understanding of user prefere-nces[C]//Proceedings of the World Wide Web Conference.2019:151-161.
[3]YU L,MADJID N A,DIFALLAH D.CrunchQA:A SyntheticDataset for Question Answering over Crunchbase Knowledge Graph[C]//2022 IEEE International Conference on Big Data.IEEE,2022:4635-4641.
[4]CHEN X,ZHANG Z,WU Q.A Survey of Meta-Learning Techniques for Knowledge Graphs[J].ACM Computing Surveys,2023,55(4):1-35.
[5]CHEN X,WU Q,LIU H.Meta-Learning for Knowledge Graph Completion:A Review[J].Knowledge-Based Systems,2023,22(3):182-193.
[6]WANG T,YANG Z,XU R.A Comprehensive Survey onKnowledge Graph Embedding Techniques[J].ACM Computing Surveys,2023,55(1):1-34.
[7]CHEN M,TIAN Y,YANG M,et al.Multilingual knowledge graph embeddings for cross-lingual knowledge alignment[C]//Proceedings of the 2018 Conference on Empirical Methods in Natural Language Processing.2016:213-225.
[8]SUN Z,HU W,LI C.Cross-lingual entity alignment via joint attribute-preserving embedding[C]//The Semantic Web-ISWC 2017:16th International Semantic Web Conference.Springer,2017:1999-2008.
[9]TROUILLON T,WELBL J,RIEDEL S,et al.Complex embeddings for simple link prediction[C]//International Conference on Machine Learning.PMLR,2016:2071-2080.
[10]ANG Z,LIU F,ZHANG Z.Cross-lingual knowledge graphalignment via graph convolutional networks[C]//Proceedings of the 2018 Conference on Empirical Methods in Natural Language Processing.2018:349-357.
[11]MAO X,LI X,LIU Y,et al.Relational reasoning for knowledge graph entity alignment[C]//Proceedings of the 2021 ACM SIGMOD International Conference on Management of Data.2021:1121-1133.
[12]ZHANG J,ZHANG W,CHEN X,et al.Multi-modal entityalignment in knowledge graphs[C]//Proceedings of the 2020 Web Conference.2020:2164-2174.
[13]WU Y,WANG X,JIANG P,et al.Knowledge-enhanced graphconvolutional networks for entity alignment[C]//Proceedings of the 29th ACM International Conference on Information and Knowledge Management.2020:1745-1754.
[14]FINN C,ABBEEL P,LEVINE S.Model-Agnostic Meta-Lear-ning for Fast Adaptation of Deep Networks[C]//Proceedings of the 34th International Conference on Machine Learning(ICML).2017:70-77.
[15]SNELL J, SWERSKY K,ZEMEL R S.Prototypical Networks for Few-Shot Learning[C]//Proceedings of the 31st International Conference on Neural Information Processing Systems.2017:30-39.
[16]YANG H W,ZOU Y,SHI P,et al.Aligning cross-lingual entities with multi-aspect information[C]//Proceedings of the 2019 Conference on Empirical Methods in Natural Language Proces-sing and the 9th International Joint Conference on Natural Language Processing(EMNLP-IJCNLP).2019:2231-2244.
[17]SANG H,CHEN W,WANG J,et al.RDGCN:Reasonably dense graph convolution network for pedestrian trajectory prediction[J].Measurement,2023,213:112675.
[18]ZHU Z,FAN X,CHU X,et al.HGCN:A heterogeneous graph convolutional network-based deep learning model toward collective classification[C]//Proceedings of the 26th ACM SIGKDD International Conference on Knowledge Discovery & Data Mi-ning.2020:1161-1171.
[19]ZHANG X,ZHANG R,CHEN J,et al.Semi-supervised entity alignment with global alignment and local information aggregation[J].IEEE Transactions on Knowledge and Data Enginee-ring,2023,35(10):10464-10477.
[20]NG Y,WU J,YU K,et al.Independent Relation Representation With Line Graph for Cross-Lingual Entity Alignment[J].IEEE Transactions on Knowledge and Data Engineering,2022,35(11):11503-11514.
[21]WANG Z,YANG J,YE X.Knowledge graph alignment with entity pair embedding[C]//Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing(EMNLP).Association for Computational Linguistics,2020:1672-1680.
[22]SUN Z,HU W,WANG C,et al.Revisiting embedding-based entity alignment:a robust and adaptive method[J].IEEE Transactions on Knowledge and Data Engineering,2022,35(8):8461-8475.
[23]WU Y,LIU X,FENG Y,et al.Neighborhood matching network for entity alignment[C]//Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics.2020:6477-6487.
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