融合思维链与知识图谱的中医问答模型

doi:10.3778/j.issn.1002-8331.2408-0101

摘要/Abstract

摘要： 针对中医问诊领域数据规模大，以及医生在问诊中主观性强、数据对齐难的问题，提出了一种中医问答领域的大语言模型ChatTCM。利用大语言模型（large language model，LLM）在处理自然语言理解与文本生成方面的强大能力，通过对大语言模型进行微调，使LLM具有在中医问答领域的专业知识和能力，避免模型在生成时出现幻觉的现象。提取中医书籍中的三元组信息，构建中医知识图谱数据库，实现中医知识的数据对齐与系统化整合，并为大语言模型生成答案提供背景知识；结合思维链（chain-of-thought，COT）与知识图谱数据库的动态交互，生成客观的推理过程，确保诊疗建议具有科学依据；把思维链与知识图谱的推理结果作为新知识进行存储，从而不断扩展本地知识库。与中医领域的HuaTuoGPT模型对比实验表明，ChatTCM模型在MedChatZH数据集上BLEU-4和ROUGE-L的评测指标分别提高了10.6和10.5个百分点，并且在已开源的数据集上准确度达到了70%，比同类型的MedChatZH模型提升了10个百分点。

关键词: 大语言模型, 微调, 知识图谱, 思维链, 中医知识

Abstract: In response to the large scale of data in the field of Chinese medicine diagnosis, as well as the high subjectivity of doctors in diagnosis and the difficulty of data alignment, ChatTCM, a large language model for the Q&A domain of traditional Chinese medicine (TCM), is proposed. Taking advantage of the power of large language model (LLM) in dealing with natural language understanding and text generation, and fine-tuning the large language model, the LLM has expertise and competence in the field of TCM Q&A, thus preventing the model from generating hallucinations. Firstly, extract triplet information from TCM books to construct a TCM knowledge graph database, achieving data alignment and systematic integration of TCM knowledge, while providing background knowledge for large language models to generate answers. Secondly, integrate chain-of-thought (COT) reasoning with dynamic interactions from the knowledge graph database to generate an objective reasoning process, ensuring that the diagnostic recommendations are based on scientific evidence. Additionally, store the reasoning results from the chain-of-thought and knowledge graph as new knowledge, continuously expanding the local knowledge base. The ChatTCM model improves the BLEU-4 and ROUGE-L metrics on the MedChatZH dataset by 10.6 and 10.5 percentage points, respectively, and achieves 70% accuracy on the open-source dataset, which is a 10 percentage points improvement over the same type of MedChatZH model.

Key words: large language model, fine-tuning, knowledge graph, chain-of-thought, knowledge of traditional Chinese medicine

苑中旭, 李理, 何凡, 杨秀, 韩东轩. 融合思维链与知识图谱的中医问答模型[J]. 计算机工程与应用, 2025, 61(4): 158-166.

YUAN Zhongxu, LI Li, HE Fan, YANG Xiu, HAN Dongxuan. Traditional Chinese Medicine Question Answering Model Based on Chain-of-Thought and Knowledge Graph[J]. Computer Engineering and Applications, 2025, 61(4): 158-166.

参考文献

[1] DEVLIN J, CHANG M W, LEE K, et al. BERT: pre-training of deep bidirectional transformers for language understanding[J]. arXiv:1810.04805, 2018.
[2] ZENG A H, LIU X, DU Z H, et al. GLM130B: an open bilingual pre-trained model[C]//Proceedings of the 2023 International Conference on Learning Representations, 2023.
[3] WANG H, LIU C, XI N, et al. HuaTuo: tuning LLaMA model with Chinese medical knowledge[J]. arXiv:2304.06975, 2023.
[4] WANG N, YANG H, WANG C D. FinGPT: instruction tuning benchmark for open-source large language models in financial datasets[J]. arXiv:2310.04793, 2023.
[5] 代佳梅, 孔韦韦, 王泽, 等. BERT和LSI的端到端方面级情感分析模型[J]. 计算机工程与应用, 2024, 60(12): 144-152.
DAI J M, KONG W W, WANG Z, et al. End-to-end aspect-based sentiment analysis model for BERT and LSI[J]. Computer Engineering and Applications, 2024, 60(12): 144-152.
[6] KRALJEVIC Z, SHEK A, BEAN D, et al. MedGPT: medical concept prediction from clinical narratives[J]. arXiv:2107.03134, 2021.
[7] MAYNEZ J, NARAYAN S, BOHNET B, et al. On faithfulness and factuality in abstractive summarization[J]. arXiv:2005.00661, 2020.
[8] SHUSTER K, POFF S, CHEN M, et al. Retrieval augmentation reduces hallucination in conversation[J]. arXiv:2104.07567, 2021.
[9] AGARWAL O, GE H, SHAKERI S, et al. Knowledge graph based synthetic corpus generation for knowledge-enhanced language model pre-training[J]. arXiv:2010.12688, 2020.
[10] 张鹤译, 王鑫, 韩立帆, 等. 大语言模型融合知识图谱的问答系统研究[J]. 计算机科学与探索, 2023, 17(10): 2377-2388.
ZHANG H Y, WANG X, HAN L F, et al. Research on question answering system on joint of knowledge graph and large language models[J]. Journal of Frontiers of Computer Science and Technology, 2023, 17(10): 2377-2388.
[11] VASWANI A, SHAZEER N, PARMAR N, et al. Attention is all you need[J]. arXiv:1706.03762, 2017.
[12] TEAM G, ANIL R, BORGEAUD S, et al. Gemini: a family of highly capable multimodal models[J]. arXiv:2312.11805, 2023.
[13] SHUSTER K, XU J, KOMEILI M, et al. BlenderBot 3: a deployed conversational agent that continually learns to responsibly engage[J]. arXiv:2208.03188, 2022.
[14] LI Y, LI Z, ZHANG K, et al. ChatDoctor: a medical chat model fine-tuned on a large language model meta-AI (LLaMA) using medical domain knowledge[J]. arXiv:2303.14070, 2023.
[15] TOUVRON H, LAVRIL T, IZACARD G, et al. LLaMA: open and efficient foundation language models[J]. arXiv:2302.13971, 2023.
[16] XIONG H, WANG S, ZHU Y, et al. DoctorGLM: fine-tuning your Chinese doctor is not a Herculean task[J]. arXiv:2304.
01097, 2023.
[17] WU C, ZHANG X, ZHANG Y, et al. PMC-LLaMA: further finetuning LLaMA on medical papers[J]. arXiv:2304.14454, 2023.
[18] WEI J, WANG X, SCHUURMANS D, et al. Chain-of-thought prompting elicits reasoning in large language models[C]//Advances in Neural Information Processing Systems 35, 2022: 24824-24837.
[19] WANG J, SUN Q, LI X, et al. Boosting language models reasoning with chain-of-knowledge prompting[J]. arXiv:2306.06427, 2023.
[20] TRIVEDI H, BALASUBRAMANIAN N, KHOT T, et al. Interleaving retrieval with chain-of-thought reasoning for knowledge-intensive multi-step questions[J]. arXiv:2212.
10509, 2022.
[21] NIU M, LI H, SHI J, et al. Mitigating hallucinations in large language models via self-refinement-enhanced knowledge retrieval[J]. arXiv:2405.06545, 2024.
[22] ZHOU T, CHEN Y, LIU K, et al. CogMG: collaborative augmentation between large language model and knowledge graph[J]. arXiv:2406.17231, 2024.
[23] WANG J, CHEN M, HU B, et al. Learning to plan for retrieval-augmented large language models from knowledge graphs[C]//Findings of the Association for Computational Linguistics: EMNLP 2024. Stroudsburg: ACL, 2024.
[24] KANG Y L, CHANG Y, FU J Y, et al. CMLM-ZhongJing: large language model is GoodStory listener[EB/OL]. [2024-07-12]. https://github.com/pariskang/CMLM-ZhongJing.
[25] 张君冬, 杨松桦. HuangDi: 中医古籍生成式大语言模型的构建研究[EB/OL]. [2024-07-12]. https://github.com/Zlasejd/HuangDi.
ZHANG J D, YANG S H. HuangDi: research on the construction of generative large language model for ancient Chinese medicine books[EB/OL]. [2024-07-12]. https://github.com/Zlasejd/HuangDi.
[26] TAN Y, LI M, HUANG Z, et al. MedChatZH: a better medical adviser learns from better instructions[J]. arXiv:2309.01114, 2023.
[27] ZHANG N, JIA Q, YIN K, et al. Conceptualized representation learning for Chinese biomedical text mining[J]. arXiv:2008.10813, 2020.
[28] 奥德玛, 杨云飞, 穗志方, 等.?中文医学知识图谱CMeKG构建初探[J].?中文信息学报, 2019, 33(10): 1-7.
ODMAA, YANG Y F, SUI Z F, et al.?Preliminary study on the construction of Chinese medical knowledge graph[J].?Journal of Chinese Information Processing, 2019, 33(10): 1-7.
[29] ROBERTSON S, ZARAGOZA H. The probabilistic relevance framework: BM25 and beyond[J]. Foundations and Trends in Information Retrieval, 2009, 3(4): 333-389.
[30] LIU X, JI K, FU Y, et al. P-tuning v2: prompt tuning can be comparable to fine-tuning universally across scales and tasks[J]. arXiv:2110.07602, 2021.
[31] HU E J, SHEN Y, WALLIS P, et al. LoRA: low-rank adaptation of large language models[J]. arXiv:2106.09685, 2021.
[32] OUYANG L, WU J, XU J, et al. Training language models to follow instructions with human feedback[C]//Advances in Neural Information Processing Systems 35, 2022: 27730-27744.
[33] ZHANG H, CHEN J, JIANG F, et al. HuatuoGPT, towards taming language model to be a doctor[C]//Findings of the Association for Computational Linguistics: EMNLP 2023. Stroudsburg: ACL, 2023: 10859-10885.
[34] WANG X, CHEN G, SONG D, et al. CMB: a comprehensive medical benchmark in Chinese[C]//Proceedings of the 2024 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies. Stroudsburg: ACL, 2024: 6184-6205.