融合大模型微调与图神经网络的知识图谱问答

doi:10.3778/j.issn.1002-8331.2406-0301

摘要/Abstract

摘要： 传统知识图谱问答系统在处理自然语言问句时，常因语义解析不精确而导致错误。为解决这一问题，提出一种融合大模型微调和图神经网络的知识图谱问答方法。收集问题并定义问题的逻辑形式；利用大型预训练语言模型的强大语义解析能力，通过对问题及其对应逻辑形式构成的问答对进行微调，提升问题解析的精度；采用模糊集方法增强微调后的逻辑形式，提高其检索精度；利用图神经网络对这些逻辑形式进行关系投影和逻辑运算获取最终答案。在通用领域标准数据集WebQSP和ComplexWebQuestions上的实验验证表明，该方法在F1、Hit@1和ACC这三个指标上均优于基准模型。同时，方法也在垂直领域风电装备数据集、高速列车数据集上进行了应用和验证。

关键词: 知识图谱问答, 大模型微调, 逻辑形式, 图神经网络检索

Abstract: To address the challenges posed by inaccurate semantic parsing in traditional knowledge graph question answering systems when processing natural language queries, this paper proposes a method that integrates large model fine-tuning with graph neural networks. The approach begins with the collection of questions and the definition of their corresponding logical forms. Leveraging the robust semantic parsing capabilities of large pre-trained language models, the accuracy of question parsing is significantly enhanced through fine-tuning on question-answer pairs, where each pair includes a question and its associated logical form. Subsequently, the fuzzy set method is applied to further refine the fine-tuned logical forms, improving retrieval precision. Finally, graph neural networks are employed to perform relation projection and logical operations on these enhanced logical forms to derive the final answers. Experimental validation on standard general-domain datasets, such as WebQSP and ComplexWebQuestions, demonstrates that this method surpasses baseline models in terms of F1, Hit@1, and ACC metrics. Additionally, the method has been successfully applied and validated on domain-specific datasets, including those related to wind power equipment and high-speed trains, confirming its effectiveness in specialized domains.

Key words: knowledge graph Q&, A, large model fine-tuning, logical form, graph neural network retrieval

陈俊臻, 王淑营, 罗浩然. 融合大模型微调与图神经网络的知识图谱问答[J]. 计算机工程与应用, 2024, 60(24): 166-176.

CHEN Junzhen, WANG Shuying, LUO Haoran. Combining Large Model Fine-Tuning and Graph Neural Networks for Knowledge Graph Question Answering[J]. Computer Engineering and Applications, 2024, 60(24): 166-176.

参考文献

[1] 曹书林, 史佳欣, 侯磊, 等. 知识库问答研究进展与展望[J]. 计算机学报, 2023, 46(3): 512-539.
CAO S L, SHI J X, HOU L, et al. Question answering over knowledge base: an overview[J]. Chinese Journal of Computers, 2023, 46(3): 512-539.
[2] 刘昀抒, 申彦明, 齐恒, 等. 基于层次结构图的多跳知识图谱问答模型[J]. 计算机工程, 2024, 50(1): 101-109.
LIU Y S, SHEN Y M, QI H, et al. Multi-hop knowledge base question answering model based on hierarchical structure graph[J]. Computer Engineering, 2024, 50(1): 101-109.
[3] SAXENA A, TRIPATHI A, TALUKDAR P. Improving multi-hop question answering over knowledge graphs using knowledge base embeddings[C]//Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, 2020: 4498-4507.
[4] CHEN Z Y, CHANG C H, CHEN Y P, et al. UHop: an unrestricted-hop relation extraction framework for knowledge-based question answering[J]. arXiv:1904.01246, 2019.
[5] LAN Y, WANG S, JIANG J. Knowledge base question answering with topic units[C]//Proceedings of the Twenty-Eighth International Joint Conference on Artificial Intelligence, 2019: 5046-5052.
[6] SHU Y, YU Z, LI Y, et al. Tiara: multi-grained retrieval for robust question answering over large knowledge bases[J]. arXiv:2210.12925, 2022.
[7] YU D, ZHANG S, NG P, et al. DecAF: joint decoding of answers and logical forms for question answering over knowledge bases[J]. arXiv:2210.00063, 2022.
[8] HU X, WU X, SHU Y, et al. Logical form generation via multi-task learning for complex question answering over knowledge bases[C]//Proceedings of the 29th International Conference on Computational Linguistics, 2022: 1687-1696.
[9] 文森, 钱力, 胡懋地, 等. 基于大语言模型的问答技术研究进展综述[J]. 数据分析与知识发现, 2024, 8(6): 16-29.
WEN S, QIAN L, HU M, et al. Review of research progress on question-answering techniques based on large language models[J]. Data Analysis and Knowledge Discovery, 2024, 8(6): 16-29.
[10] BROWN T B. Language models are few-shot learners[J]. arXiv:2005.14165, 2020.
[11] TOUVRON H, LAVRIL T, IZACARD G, et al. LLaMA: open and efficient foundation language models[J]. arXiv:2302.13971, 2023.
[12] YASUNAGA M, REN H, BOSSELUT A, et al. QA-GNN: reasoning with language models and knowledge graphs for question answering[J]. arXiv:2104.06378, 2021.
[13] LUO H R, E H H, TANG Z C, et al. ChatKBQA: a generate-then-retrieve framework for knowledge base question answering with fine-tuned large language models[J]. arXiv:2310.08975, 2023.
[14] BHUTANI N, ZHENG X, JAGADISH H V. Learning to answer complex questions over knowledge bases with query composition[C]//Proceedings of the 28th ACM International Conference on Information and Knowledge Management, 2019: 739-748.
[15] 冯钧, 李艳, 杭婷婷. 问答系统中复杂问题分解方法研究综述[J]. 计算机工程与应用, 2022, 58(17): 23-33.
FENG J, LI Y, HANG T T. Survey on question decomposition method in question answering system[J]. Computer Engineering and Applications, 2022, 58(17): 23-33.
[16] 萨日娜, 李艳玲, 林民. 知识图谱推理问答研究综述[J]. 计算机科学与探索, 2022, 16(8): 1727-1741.
SA R N, LI Y L, LIN M. Survey of question answering based on knowledge graph reasoning[J]. Journal of Frontiers of Computer Science and Technology, 2022, 16(8): 1727-1741.
[17] GARCíA-DURáN A, DUMAN?I? S, NIEPERT M. Learning sequence encoders for temporal knowledge graph completion[J]. arXiv:1809.03202, 2018.
[18] CHEN X, HU Z, SUN Y. Fuzzy logic based logical query answering on knowledge graphs[C]//Proceedings of the AAAI Conference on Artificial Intelligence, 2022: 3939-3948.
[19] 刘杰, 尚学群, 宋凌云, 等. 图神经网络在复杂图挖掘上的研究进展[J]. 软件学报, 2022, 33(10): 3582-3618.
LIU J, SHANG X Q, SONG L Y, et al. Progress of graph neural networks in complex graph mining[J]. Journal of Software, 2022, 33(10): 3582-3618.
[20] VASHISHTH S, SANYAL S, NITIN V, et al. Composition-based multi-relational graph convolutional networks[J]. arXiv:1911.03082, 2019.
[21] REN H, HU W, LESKOVEC J. Query2box: reasoning over knowledge graphs in vector space using box embeddings[J]. arXiv:2002.05969, 2020.
[22] HU Z, WANG L, LAN Y, et al. LLM-adapters: an adapter family for parameter-efficient fine-tuning of large language models[J]. arXiv:2304.01933, 2023.
[23] GOTTWALD S. Fuzzy sets and fuzzy logic: the foundations of application from a mathematical point of view[M]. [S.l.]: Springer-Verlag, 2013.
[24] ZHU Z, GALKIN M, ZHANG Z, et al. Neural-symbolic models for logical queries on knowledge graphs[C]//Proceedings of the International Conference on Machine Learning, 2022: 27454-27478.
[25] BOLLACKER K, EVANS C, PARITOSH P, et al. Freebase: a collaboratively created graph database for structuring human knowledge[C]//Proceedings of the 2008 ACM SIGMOD International Conference on Management of Data, 2008: 1247-1250.