2023年 03期

基于三元组键值对超关系学习的知识图谱链接预测

Link Prediction of Knowledge Graph Based on Triple Key-Value Pair Hyper-relation Learning


摘要(Abstract):

为了提高挖掘超关系数据中的隐含关系的能力,提出一种基于三元组键值对超关系学习的知识图谱链接预测模型;所提模型不仅捕捉编码在三元组中知识图谱的主要结构信息,而且通过对三元组及其相关键值对进行学习,采集每个三元组及其相关联的键值对之间的相关性;通过最小值操作,对基本三元组、键值对以及相关性特征向量进行合并,利用全连接投影得到预测得分。结果表明,与其他预测模型相比,所提模型在JF17K数据集和WikiPeople数据集的键、值预测平均倒数排名表现更佳,在链接预测中排名分别小于10、1的三元组的平均占比的预测性能更优。

关键词(KeyWords):知识图谱;链接预测;超关系学习;结构信息;键值对

基金项目(Foundation):国家自然科学基金项目(61702026);; 湖南省教育厅科学研究项目优秀青年项目(19B397)

作者(Author): 甄春成,王琼,龚芝,李庆珍,周湘贞

DOI: 10.13349/j.cnki.jdxbn.20230329.003

参考文献(References):

[1] 李浩,张亚钏,康雁,等.融合循环知识图谱和协同过滤电影推荐算法[J].计算机工程与应用,2020,56(2):106.

[2] UYAR A,ALIYU F M.Evaluating search features of Google Knowledge Graph and Bing Satori:entity types,list searches and query interfaces[J].Online Information Review,2015,39(2):197.

[3] TURKI H,VRANDECIC D ,HAMDI H ,et al.Using WikiData as a multi-lingual multi-dialectal dictionary for Arabic dialects[C]//2017 IEEE/ACS 14th International Conference on Computer Systems and Applications (AICCSA),October 30-November 3,2017,Hammamet,Tunisia.New York:IEEE,2017:75.

[4] 覃晓,元昌安,何国对,等.基于百度百科知识图谱的旅游资源文化特色评分算法:CN110781300A[P].2020-02-11.

[5] XIONG C Y,POWER R,CALLAN J.Explicit semantic ranking for academic search via knowledge graph embedding[C]//WWW’17:Proceedings of the 26th International Conference on World Wide Web,April 3-7,2017,Perth,Australia.New York:ACM,2017:1271.

[6] YANG D Q,ROSSO P,LI B,et al.NodeSketch:highly-efficient graph embeddings via recursive sketching[C]// KDD’19:Proceedings of the 25th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining,August 4-8,2019,Anchorage,USA.New York:ACM,2019:1162.

[7] 苏荣.基于图网络的知识图谱嵌入研究[D].哈尔滨:哈尔滨工业大学,2019.

[8] BEIERLE C,EICHHORN C,KERN-ISBERNER G,et al.Pro-perties of skeptical c-inference for conditional knowledge bases and its realization as a constraint satisfaction problem[J].Annals of Mathematics and Artificial Intelligence,2018,83(3/4):247.

[9] BEAN D M,WU H H,IQBAL E,et al.Author correction:knowledge graph prediction of unknown adverse drug reactions and validation in electronic health records[J].Scientific Reports,2018,8(1):4284.

[10] ZHANG R C,LI J P,MEI J J,et al.Scalable instance reconstruction in knowledge bases via relatedness affiliated embedding[C]// WWW ’18:Proceedings of the Web Conference 2018,April 23-27,2018,Lyon,France.New York:ACM,2018:1185.

[11] 邹鼎杰.基于知识图谱和贝叶斯分类器的图书分类[J].计算机工程与设计,2020,41(6):1796.

[12] ABEDINI F,KEYVANPOUR M R,MENHAJ M B.Correction tower:a general embedding method of the error recognition for the knowledge graph correction[J].International Journal of Pattern Recognition and Artificial Intelligence,2020,34(10):2059034.

[13] 刘琼昕,马敬,郑培雄.一种融合实体关联性约束的表示学习方法[J].北京理工大学学报,2020,40(1):90.

[14] GUAN S P,JIN X L,WANG Y Z,et al.Link prediction on n-ary relational data[C]//WWW ’19:the World Wide Web Conference,May 13-17,2019,San Francisco,USA.New York:ACM,2019:583.

[15] CAI H Y,ZHENG V W,CHANG K C.A comprehensive survey of graph embedding:problems,techniques,and applications[J].IEEE Transactions on Knowledge and Data Engineering,2018,30(9):1616.

[16] 王琳,张国坤,徐逸浓.一种张量分解的知识图谱推理优化算法[J].计算机应用研究,2020,37(增刊1):52.

[17] BERRENDORF M,FAERMAN E,MELNYCHUK V,et al.Knowledge graph entity alignment with graph convolutional networks:lessons learned[M]//JOSE J M,YILMAZ E,MAGALH?ES J,et al.Advances in Information Retrieval.Cham:Springer,Cham,2020:3.

[18] FANG W,ZHANG J W,WANG D L,et al.Entity disambiguation by knowledge and text jointly embedding[C]//Proceedings of the 20th SIGNLL Conference on Computational Natural Language Learning(CoNLL),August 11-12,2016,Berlin,Germany.Berlin:ACL,2016:260.