基于规则和嵌入联合学习的低资源知识图谱补全

doi:10.3778/j.issn.1002-8331.2401-0012

摘要/Abstract

摘要： 知识图谱补全（knowledge graph completion，KGC）是一种通过关系提取来填补缺失链接的知识图谱改进方法，规则挖掘和嵌入学习都可以用于知识图谱的补全。现有规则挖掘的研究主要集中在富资源语言的规则挖掘方面，在低资源知识图谱中难以挖掘出合适的规则；现有嵌入学习的方法依赖学习大量的事实三元组，难于处理低资源知识图谱的补全问题。针对两种方法存在的问题，提出一种基于规则和嵌入联合学习的低资源知识图谱补全方法。它由四部分组成，关系分流策略将三元组分为高频关系三元组和低频关系三元组；基于关系注意力的规则挖掘模块，利用改进的关系注意力机制提取低频关系的上下文信息，将其作为额外信息进行规则的挖掘；嵌入学习的模块，利用嵌入学习对知识图谱中高频关系三元组进行补全；迭代学习模块将两种方法进行有效结合，使得两种方法分别处理自己擅长的数据。通过在低资语言知识图谱补全数据集EL（希腊语）、FR（法语）、JA（日语）、Vi-310（越南语）中进行实验，使用指标MMR、HIT@1、HIT@5、HIT@10进行评估，取得了更佳的表现。

关键词: 知识图谱补全, 嵌入学习, 知识图谱规则挖掘, 迭代学习, 低资源

Abstract: Knowledge graph completion(KGC) is a knowledge graph improvement method that fills missing links through relationship extraction, and both rule mining and embedding learning can be used for knowledge graph completion. Existing research on rule mining mainly focuses on rule mining for resource-rich languages, and it is difficult to mine suitable rules in low-resource knowledge graphs; existing embedding learning methods rely on learning a large number of fact triples, which is difficult to deal with the problem of low-resource knowledge graph completion. To address this problem, a rule-based and embedded joint learning for low-resource knowledge graph completion method is proposed. It consists of four parts, the proposed relational triage strategy divides the ternary into high-frequency relationship ternary and low-frequency relationship ternary, the relational attention-based rule mining module, which extracts the contextual information of low-frequency relationships using the improved relational attention mechanism, and uses it as additional information for rule mining; the module of embedded learning, which utilizes embedded learning to complement the ternary of high-frequency relationships in the knowledge graph; and the iterative learning module combines the two methods effectively, making the two methods deal with the data they specialize in respectively. Better performance is achieved by conducting experiments in the low-resource language knowledge graph complementation datasets EL(Greek), FR(French), JA(Japanese) and Vi-310(Vietnamese), which are evaluated using the metrics MMR, HIT@1, HIT@5, and HIT@10.

Key words: knowledge graph completion, embedding learning, knowledge graph rule mining, iterative learning, low-resource

周家啟, 宋燃, 余正涛, 黄于欣, 徐佳龙. 基于规则和嵌入联合学习的低资源知识图谱补全[J]. 计算机工程与应用, 2025, 61(9): 148-158.

ZHOU Jiaqi, SONG Ran, YU Zhengtao, HUANG Yuxin, XU Jialong. Rule-Based and Embedded Joint Learning for Low-Resource Knowledge Graph Completion[J]. Computer Engineering and Applications, 2025, 61(9): 148-158.

参考文献

[1] WANG X, HE X, CAO Y, et al. AliCG: fine-grained and evolvable conceptual graph construction for semantic search at Alibaba[C]//Proceedings of the 27th ACM SIGKDD Conference on Knowledge Discovery & Data Mining, 2021: 950-958.
[2] WANG X, HE X, CAO Y, et al. KGAT: knowledge graph attention network for recommendation[C]//Proceedings of the 25th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining, 2019: 950-958.
[3] QIU Y, WANG Y, JIN X, et al. Stepwise reasoning for multi-relation question answering over knowledge graph with weak supervision[C]//Proceedings of the 13th International Conference on Web Search and Data Mining, 2020:474-482.
[4] BORDES A, USUNIER N, GARCIA-DURAN A, et al. Translating embeddings for modeling multi-relational data[C]//Advances in Neural Information Processing Systems, 2013: 2787-2795.
[5] 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.
[6] SUN Z, DENG Z H, NIE J Y, et al. Rotate: knowledge graph embedding by relational rotation in complex space[J]. arXiv:1902.10197, 2019.
[7] YANG F, YANG Z, COHEN W W. Differentiable learning of logical rules for knowledge base reasoning[C]//Advances in Neural Information Processing Systems, 2017: 2319-2328.
[8] SHEN Y, CHEN J, HUANG P S, et al. M-walk: learning to walk over graphs using monte carlo tree search[C]//Advances in Neural Information Processing Systems, 2018: 6786-6797.
[9] HO V T, Stepanova D, GAD-ELRAB M H, et al. Rule learning from knowledge graphs guided by embedding models[C]// Proceedings of the 17th International Semantic Web Conference on Semantic Web(ISWC 2018), 2018: 72-90.
[10] LIN Y, LIU Z, SUN M, et al. Learning entity and relation embeddings for knowledge graph completion[C]//Proceedings of the AAAI Conference on Artificial Intelligence, 2015.
[11] TROUILLON T, WELBL J, RIEDEL S, et al. Complex embeddings for simple link prediction[C]//Proceedings of the International Conference on Machine Learning, 2016:2071-2080.
[12] COHEN W W. TensorLog: a differentiable deductive database[J]. arXiv:1605.06523, 2016.
[13] WANG P W, STEPANOVA D, DOMOKOS C, et al. Differentiable learning of numerical rules in knowledge graphs[C]//Proceedings of the International Conference on Learning Representations, 2019: 1-12.
[14] GUO S, WANG Q, WANG L, et al. Knowledge graph embedding with iterative guidance from soft rules[C]//Proceedings of the AAAI Conference on Artificial Intelligence, 2018: 4815-4823.
[15] XU Z, YE P, CHEN H, et al. Ruleformer: context-aware rule mining over knowledge graph[C]//Proceedings of the 29th International Conference on Computational Linguistics, 2022: 2551-2560.
[16] GENG Y, CHEN J, PAN J Z, et al. Relational message passing for fully inductive knowledge graph completion[C]//Proceedings of the 2023 IEEE 39th International Conference on Data Engineering (ICDE), 2023: 1221-1233.
[17] ZHANG K, ZHANG D, LIU N, et al. Sample and feature enhanced few-shot knowledge graph completion[C]//Proceedings of the International Conference on Database Systems for Advanced Applications. Cham: Springer Nature Switzerland, 2023: 800-809.
[18] ZHANG W, ?ERBAN O, SUN J, et al. Conflict-aware multilingual knowledge graph completion[J]. Knowledge-Based Systems, 2023, 281: 111070.
[19] OMELIYANENKO J, ZEHE A, HOTHO A, et al. CapsKG: enabling continual knowledge integration in language models for automatic knowledge graph completion[C]// Proceedings of the International Semantic Web Conference. Cham: Springer Nature Switzerland, 2023: 618-636.
[20] ZHANG C, YAO H, HUANG C, et al. Few-shot knowledge graph completion[C]//Proceedings of the AAAI Conference on Artificial Intelligence, 2020: 3041-3048.
[21] KAZEMI S M, POOLE D. Simple embedding for link prediction in knowledge graphs[C]//Advances in Neural Information Processing Systems, 2018: 4284-4295.
[22] NIU G, LI Y, TANG C, et al. Relational learning with gated and attentive neighbor aggregator for few-shot knowledge graph completion[C]//Proceedings of the 44th International ACM SIGIR Conference on Research and Development in Information Retrieval, 2021: 213-222.
[23] CHEN M, ZHANG W, ZHANG W, et al. Meta relational learning for few-shot link prediction in knowledge graphs[J]. arXiv:1909.01515, 2019.
[24] PEI S, KOU Z, ZHANG Q, et al. Few-shot low-resource knowledge graph completion with multi-view task representation generation[C]//Proceedings of the 29th ACM SIGKDD Conference on Knowledge Discovery and Data Mining, 2023: 1862-1871.
[25] SONG R, HE S, GAO S, et al. Multilingual knowledge graph completion from pretrained language models with knowledge constraints[C]//Findings of the Association for Computational Linguistics(ACL 2023), 2023: 7709-7721.