基于邻域聚合与CNN的知识图谱实体类型补全

doi:10.19678/j.issn.1000-3428.0063989

摘要/Abstract

摘要： 现有知识图谱实体类型补全模型通过对实体和实体类型进行建模，以解决知识图谱中实体缺失的实体类型，但未有效地利用实体之间的关系，导致模型的实体类型补全性能不佳。提出一种基于邻域聚合与卷积神经网络的知识图谱实体类型补全模型NACE2T，其采用编码器-解码器的结构。基于注意力机制设计利用关系信息的编码器，其使用注意力机制为实体邻域中的每个关系-实体对分配不同的权重，以聚合实体邻域中实体和关系的信息，从而利用实体之间的关系。基于卷积神经网络设计一个新的知识图谱实体类型补全模型CE2T，将其作为解码器，对编码器输出的实体嵌入和实体类型嵌入进行建模与实体类型补全。实验结果表明，相比ConnectE模型，NACE2T模型在数据集FB15KET上的HITS@1和HITS@3提高约1.5%，在数据集YAGO43KET上的MRR和HITS@3提高约6%，HITS@1提高约9%，能够有效地推断知识图谱中实体缺失的实体类型。

关键词: 知识图谱, 实体类型, 邻域聚合, 注意力机制, 关系-实体对, 卷积神经网络

Abstract: Existing entity type completion models of the knowledge graph solve the entity types missing in the knowledge graph by modeling entities and entity types.However, they do not effectively use the relationships between entities, which results in the poor performance of the entity type completion of the model.An entity type completion model, NACE2T, of a knowledge graph based on neighborhood aggregation and Convolutional Neural Network(CNN) is proposed, which adopts the encoder-decoder structure.Based on the attention mechanism, an encoder is designed to use the relationship information.The attention mechanism is used to assign different weights to each relationship-entity pair in the entity neighborhood, and it is used to aggregate the information of entities and relationships in the entity neighborhood to use the relationship between entities.Based on CNN, a new entity type completion model called CE2T of knowledge graph is designed.This model is used as a decoder to model and complete entity types through entity embedding and entity type embedding of the encoder output.The experimental results show that compared with the ConnectE model, the NACE2T model HITS@1 and HITS@3 increase by approximately 1.5% on FB15KET dataset, MRR and HITS@3 increase by approximately 6%, and HITS@1 increase by approximately 9% on YAGO43KET dataset.The proposed model can effectively infer the types of entities missing in the knowledge graph.

Key words: knowledge graph, entity type, neighborhood aggregation, attention mechanism, relationship-entity pair, Convolutional Neural Network(CNN)

中图分类号:

TP391

邹长龙, 安敬民, 李冠宇. 基于邻域聚合与CNN的知识图谱实体类型补全[J]. 计算机工程, 2023, 49(3): 134-141.

ZOU Changlong, AN Jingmin, LI Guanyu. Knowledge Graph Entity Type Completion Based on Neighborhood Aggregation and CNN[J]. Computer Engineering, 2023, 49(3): 134-141.

https://www.ecice06.com/CN/Y2023/V49/I3/134

图/表 9

20230314185310

20230314185313

20230314185316

20230314185319

20230314185324

20230314185330

20230314185333

20230314185337

20230314185340

参考文献

[1] 舒世泰, 李松, 郝晓红, 等.知识图谱嵌入技术研究进展[J].计算机科学与探索, 2021, 15(11):2048-2062. SHU S T, LI S, HOU X H, et al.Knowledge graph embedding technology:a review[J].Journal of Frontiers of Computer Science & Technology, 2021, 15(11):2048-2062.(in Chinese)
[2] ZHAO Y, LI Z Q, DENG W, et al.Learning entity type structured embeddings with trustworthiness on noisy knowledge graphs[J].Knowledge-Based Systems, 2021, 215:1-10.
[3] ZHU Q, WEI H, SISMAN B, et al.Collective multi-type entity alignment between knowledge graphs[C]//Proceedings of the Web Conference.New York, USA:ACM Press, 2020:2241-2252.
[4] CHEN S, WANG J P, JIANG F, et al.Improving entity linking by modeling latent entity type information[C]//Proceedings of the AAAI Conference on Artificial Intelligence.[S.l.]:AAAI Press, 2020, 34(5):7529-7537.
[5] NIU G L, LI B, ZHANG Y F, et al.AutoETER:automated entity type representation for knowledge graph embedding[C]//Proceedings of Findings of the Association for Computational Linguistics.Stroudsburg, USA:Association for Computational Linguistics, 2020:1172-1181.
[6] 金婧, 万怀宇, 林友芳.融合实体类别信息的知识图谱表示学习[J].计算机工程, 2021, 47(4):77-83. JIN J, WAN H Y, LIN Y F.Knowledge graph representation learning fused with entity category information[J].Computer Engineering, 2021, 47(4):77-83.(in Chinese)
[7] BISWAS R, SOFRONOVA R, ALAM M, et al.Entity type prediction in knowledge graphs using embeddings[EB/OL].[2022-01-18].https://arxiv.org/abs/2004.13702v2.
[8] TEACH Y C, RUFFINELLI D, BROSCHEIT S, et al.You CAN teach an old dog new tricks! on training knowledge graph embeddings[C]//Proceedings of International Conference on Learning Representations.Washington D.C., USA:[s.n.], 2020:1-10.
[9] ZHANG Q, ZHANG L, QIN C, et al.A survey on knowledge graph-based recommender systems[J].Scientia Sinica Informationis, 2020, 50(7):937-956.
[10] MOON C, JONES P, SAMATOVA N F.Learning entity type embeddings for knowledge graph completion[C]//Proceedings of Conference on Information and Knowledge Management.New York, USA:ACM Press, 2017:2215-2218.
[11] ZHAO Y, ZHANG A X, XIE R B, et al.Connecting embeddings for knowledge graph entity typing[C]//Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics.Stroudsburg, USA:Association for Computational Linguistics, 2020:6419-6428.
[12] DETTMERS T, MINERVINI P, STENETORP P, et al.Convolutional 2D knowledge graph embeddings[C]//Proceedings of the AAAI Conference on Artificial Intelligence.[S.l.]:AAAI Press, 2018:1811-1818.
[13] MONIRUZZAMAN A B M, NAYAK R, TANG M L, et al.Fine-grained type inference in knowledge graphs via probabilistic and tensor factorization methods[C]//Proceedings of World Wide Web Conference.New York, USA:ACM Press, 2019:3093-3100.
[14] LI Q.A neural model for type classification of entities for text[J].Knowledge-Based Systems, 2019, 176:122-132.
[15] BORDES A, USUNIER N, GARCÍADURÁN A, et al.Translating embeddings for modeling multi-relational data[C]//Proceedings of Conference on Neural Information Processing Systems.[S.l.]:NIPS Press, 2013:2787-2795.
[16] NICKEL M, TRESP V, KRIEGEL H P.A three-way model for collective learning on multi-relational data[C]//Proceedings of the 28th International Conference on Machine Learning.Washington D.C., USA:IEEE Press, 2011:809-816.
[17] NICKEL M, ROSASCO L, POGGIO T.Holographic embeddings of knowledge graphs[C]//Proceedings of the AAAI Conference on Artificial Intelligence.[S.l.]:AAAI Press, 2016:1955-1961.
[18] MOON C S, HARENBERG S, SLANKAS J, et al.Learning contextual embeddings for knowledge graph completion[EB/OL].[2022-01-18].https://www.researchgate.net/profile/Changsung-Moon/publication/318471812_Learning_Contextual_Embeddings_for_Knowledge_Graph_Completion/links/596d4ce0458515d9265fc8b6/Learning-Contextual-Embeddings-for-Knowledge-Graph-Completion.pdf.
[19] JI G L, HE S Z, XU L H, et al.Knowledge graph embedding via dynamic mapping matrix[C]//Proceedings of the 53rd Annual Meeting of the Association for Computational Linguistics.Stroudsburg, USA:Association for Computational Linguistics, 2015:687-696.
[20] VASHISHTH S, SANYAL S, NITIN V, et al.Composition-based multi-relational graph convolutional networks[C]//Proceedings of the 8th International Conference on Learning Representations.Washington D.C., USA:[s.n.], 2020:1-15.
[21] KIPF T, WELLING M.Semi-supervised classification with graph convolutional networks[EB/OL].[2022-01-18].https://arxiv.org/pdf/1609.02907.pdf.
[22] SHANG C, TANG Y, HUANG J, et al.End-to-end structure-aware convolutional networks for knowledge base completion[EB/OL].[2022-01-18].https://arxiv.org/pdf/1811.04441.pdf.
[23] ZHANG S, RAO X, TAY Y, et al.Knowledge router:learning disentangled representations for knowledge graphs[C]//Proceedings of Conference on North American Chapter of the Association for Computational Linguistics:Human Language Technologies.Stroudsburg, USA:Association for Computational Linguistics, 2021:1-6.
[24] NGUYEN D Q, NGUYEN T, NGUYEN D Q, et al.A novel embedding model for knowledge base completion based on convolutional neural network[EB/OL].[2022-01-18].https://arxiv.org/pdf/1712.02121.pdf.
[25] ROSSO P, YANG D Q, CUDRÉ-MAUROUX P.Beyond triplets:hyper-relational knowledge graph embedding for link prediction[C]//Proceedings of the Web Conference. New York, USA:ACM Press, 2020:1885-1896.
[26] NGUYEN D Q, VU T, NGUYEN T D, et al.A capsule network-based embedding model for knowledge graph completion and search personalization[EB/OL].[2022-01-18].https://arxiv.org/pdf/1808.04122.pdf.

选择文件类型/文献管理软件名称

选择包含的内容