提高长尾数据知识图谱补全性能的一种新算法

doi:10.15960/j.cnki.issn.1007-6093.2025.01.004

摘要/Abstract

摘要：

知识图谱是众多智能应用中一种重要的语义数据, 但其数据的不完备性给实际应用带来了很多困难, 因此需要对知识图谱中缺失的语义信息进行补全。知识图谱嵌入是知识图谱补全的重要方法之一, 这类方法通常在非长尾数据情况下具有较好的效果, 但在长尾数据情况下其效果较差。由于非长尾数据的语义较丰富, 为了提升长尾数据情况下知识图谱补全效果, 本文将非长尾数据作为监督知识迁移到长尾数据中, 提出了一种新的算法——融入期望最大化算法思想的双重嵌入方法, 来改进长尾数据的知识图谱补全性能, 进而提高其实际应用效果。通过在FB15K数据集中进行链接预测任务的对比实验, 实验结果表明本文提出的融入期望最大化算法思想的双重嵌入方法效果较好。

关键词: 知识图谱补全, 知识图谱嵌入, 期望最大化算法, 双重嵌入方法

Abstract:

Knowledge graph is an important semantic data in many intelligent applications, but the incompleteness of its data brings many difficulties to practical applications. Therefore, it is necessary to complete the missing semantic information in the knowledge graph. Knowledge graph embedding is one of the important methods of knowledge graph completion. This type of method usually has better results in the case of non-long-tail data, but its effect is poor in the case of long-tail data. Since the semantics of non-long-tail data is richer, in order to improve the effect of knowledge graph completion in the case of long-tail data, this paper transfers non-long-tail data as supervised knowledge to long-tail data, and proposes a new algorithm——the dual embedding method with the idea of expectation maximization algorithm, to improve the completion performance of knowledge graph, and its practical application effect. Through the comparative experiment of link prediction task in FB15K data set, the experimental results show that the dual embedding method with the idea of expectation maximization algorithm proposed in this paper is effective.

Key words: knowledge graph completion, knowledge graph embedding, expectation maximization algorithm, dual embedding meth

中图分类号:

O224
TP181

何苗惠, 段旭祥, 吴至友. 提高长尾数据知识图谱补全性能的一种新算法[J]. 运筹学学报（中英文）, 2025, 29(1): 41-54.

Miaohui HE, Xuxiang DUAN, Zhiyou WU. A new algorithm for improving the completion performance of knowledge graph of long-tail data[J]. Operations Research Transactions, 2025, 29(1): 41-54.

图/表 8

图1

图2

表1

表2

表3

表4

表5

表6

参考文献 23

1	Yang Y, Agrawal D, Jagadish H V, et al. An efficient parallel keyword search engine on knowledge graphs [C]//2019 IEEE 35th International Conference on Data Engineering, 2019: 338-349.
2	Yao X, Van Durme B. Information extraction over structured data: Question answering with freebase [C]//Proceedings of the 52nd Annual Meeting of the Association for Computational Linguistics, 2014: 956-966.
3	Zhang F, Yuan N J, Lian D, et al. Collaborative knowledge base embedding for recommender systems [C]//Proceedings of the 22nd ACM SIGKDD Iinternational Conference on Knowledge Discovery and Data Mining, 2016: 353-362.
4	Rotmensch M , Halpern Y , Tlimat A , et al. Learning a health knowledge graph from electronic medical records[J]. Scientific Reports, 2017, 7 (1): 1- 11. doi: 10.1038/s41598-016-0028-x
5	Suchanek F M , Kasneci G , Weikum G . Yago: A large ontology from wikipedia and wordnet[J]. Journal of Web Semantics, 2008, 6 (3): 203- 217. doi: 10.1016/j.websem.2008.06.001
6	Lehmann J , Isele R , Jakob M , et al. DBpedia\|A large-scale, multilingual knowledge base extracted from wikipedia[J]. Semantic Web, 2015, 6 (2): 167- 195. doi: 10.3233/SW-140134
7	Miller G A . WordNet: A lexical database for English[J]. Communications of the ACM, 1995, 38 (11): 39- 41. doi: 10.1145/219717.219748
8	Carlson A, Betteridge J, Kisiel B, et al. Toward an architecture for never-ending language learning [C]//Twenty-fourth AAAI Conference on Artificial Intelligence, 2010: 1306-1313.
9	Bordes A , Usunier N , Garcia-Duran A , et al. Translating embeddings for modeling multirelational data[J]. Advances in Neural Information Pprocessing Systems, 2013, 2, 2787- 2795.
10	Wang Z, Zhang J, Feng J, et al. Knowledge graph embedding by translating on hyperplanes [C]//Proceedings of the AAAI Conference on Artificial Intelligence, 2014, 28(1): 1112-1119.
11	Lin Y, Liu Z, Sun M, et al. Learning entity and relation embeddings for knowledge graph completion [C]//Twenty-ninth AAAI Conference on Artificial Intelligence, 2015: 2181-2187.
12	Ji G, He S, Xu L, et al. Knowledge graph embedding via dynamic mapping matrix [C]//Proceedings of the 53rd Annual Meeting of the Association for Computational Linguistics and the 7th International Joint Conference on Natural Language Processing, 2015: 687-696.
13	Nickel M, Tresp V, Kriegel H P. A three-way model for collective learning on multi-relational data [C]//International Conference on Machine Learning, 2011: 809-816.
14	Yang B, Yih W, He X, et al. Embedding entities and relations for learning and inference in knowledge bases [EB/OL]. (2015-08-29)[2021-11-26]. arXiv: 1412.6575.
15	Trouillon T, Welbl J, Riedel S, et al. Complex embeddings for simple link prediction [C]//International Conference on Machine Learning, 2016: 2071-2080.
16	Dettmers T, Minervini P, Stenetorp P, et al. Convolutional 2nd knowledge graph embeddings [C]//Thirty-second AAAI Conference on Artificial Intelligence, 2018: 1811-1818.
17	Vu T, Nguyen T D, Nguyen D Q, et al. A capsule network-based embedding model for knowledge graph completion and search personalization [C]//Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, 2019: 2180-2189.
18	Sun Z Q, Deng Z H, Nie J Y, et al. RotatE: Knowledge graph embedding by relational rotation in complex space [C]//Proceedings of the International Conference on Learning Representations, 2019: 926-934.
19	Zhang S, Tay Y, Yao L, et al. Quaternion knowledge graph embeddings [C]//Advances in Neural Information Processing Systems, 2019: 2731-2741.
20	Akrami F, Saeef M S, Zhang Q, et al. Realistic re-evaluation of knowledge graph completion methods: An experimental study [C]//Proceedings of the 2020 ACM SIGMOD International Conference on Management of Data, 2020: 1995-2010.
21	Metz L, Maheswaranathan N, Cheung B, et al. Meta-learning update rules for unsupervised representation learning [EB/OL]. (2019-02-26)[2021-11-25]. arXiv: 1804.00222.
22	Zhang Z , Zhuang F , Qu M , et al. Knowledge graph embedding with shared latent semantic units[J]. Neural Networks, 2021, 139, 140- 148. doi: 10.1016/j.neunet.2021.02.013
23	Dempster A P , Laird N M , Rubin D B . Maximum likelihood from incomplete data via the EM algorithm[J]. Journal of the Royal Statistical Society: Series B (Methodological), 1977, 39 (1): 1- 22. doi: 10.1111/j.2517-6161.1977.tb01600.x

	$ \alpha _1 $	$ \alpha _2 $	$ T $	$ \gamma $	neg	$ \left\| {{{{\cal L}}^e}} \right\| $	$ \left\| {{{{\cal L}}^r}} \right\| $
TransE	0.3	$ - $	300	3	25	$ - $	$ - $
TransE-DEM	$ - $	0.000 5	300	3	25	300	200
TransH	0.3	$ - $	300	3	25	$ - $	$ - $
TransH-DEM	$ - $	0.000 5	300	3	25	300	200
TransD	0.3	$ - $	300	3	25	$ - $	$ - $
TransD-DEM	$ - $	0.000 5	300	3	25	300	200

冗余关系	反向关系	笛卡尔积关系	近似冗余关系
数量	429 543	983	29 288
百分比/%	88.9	0.2	6.0

	训练集			测试集
	非长尾数据	长尾数据	总数	非长尾数据	长尾数据	总数
数量	438 364	44 778	483 142	53 649	5 422	59 071

	MR	FMR	MRR	FMRR	Hit@10	FHit@10
TransE	224.81	129.40	0.248	0.418	0.498	0.658
TransE-DEM	$\underline{\textbf{201.88}}$	$\underline{\textbf{105.03}}$	$\textbf{0.256}$	$\textbf{0.425}$	$\textbf{0.502}$	$\textbf{0.670}$
TransH	221.45	126.64	0.245	0.413	0.495	0.657
TransH-DEM	$\textbf{218.20}$	$\textbf{121.34}$	$\underline{\textbf{0.259}}$	$\underline{\textbf{0.434}}$	$\underline{\textbf{0.506}}$	$\underline{\textbf{0.675}}$
TransD	221.82	126.62	0.246	0.417	0.497	0.660
TransD-DEM	$\textbf{209.84}$	$\textbf{113.41}$	$\textbf{0.258}$	$\textbf{0.430}$	$\textbf{0.505}$	$\textbf{0.670}$

	MR	FMR	MRR	FMRR	Hit@10	FHit@10
TransE	203.17	119.80	0.250	0.427	0.515	0.674
TransE-DEM	$\underline{\textbf{180.61}}$	$\underline{\textbf{97.79}}$	$\textbf{0.260}$	$\textbf{0.433}$	$\textbf{0.519}$	$\textbf{0.683}$
TransH	200.73	117.56	0.246	0.422	0.511	0.671
TransH-DEM	$\textbf{195.94}$	$\textbf{112.51}$	$\underline{\textbf{0.262}}$	$\underline{\textbf{0.441}}$	$\underline{\textbf{0.523}}$	$\underline{\textbf{0.688}}$
TransD	199.81	116.63	0.249	0.427	0.514	0.676
TransD-DEM	$\textbf{188.55}$	$\textbf{105.68}$	$\textbf{0.260}$	$\textbf{0.436}$	$\textbf{0.522}$	$\textbf{0.682}$