基于多文档异构结构图的检索增强生成方法
Retrieval-Augmented Generation Based on Multi-Document Heterogeneous Structure Graphs
DOI: 10.12677/csa.2026.169298, PDF,   
作者: 师培仁：河北地质大学信息工程学院，河北 石家庄；柴变芳：河北省地质环境感知与数据处理重点实验室，河北 石家庄
关键词: GraphRAG；多文档问答；异构文档；层级图索引；GraphRAG； Multi-Document Question Answering； Heterogeneous Documents； Hierarchical Graph Indexing
摘要: 针对财政制度多文档问答任务中证据分散、结构异构和生成依据不足等问题，提出一种基于多文档异构结构图的检索增强生成方法(Multi-Document Heterogeneous Structural Graph-Based Retrieval-Augmented Generation Method, MDHSG-RAG)。该方法通过结构模板构建统一层级图索引，按问题类型进行由粗到细地分层检索，并以证据链约束答案生成。实验结果表明，MDHSG-RAG在检索和生成阶段均优于对比方法，提高了财政制度问答的证据定位能力、答案准确性和可追溯性。
Abstract: To address the problems of scattered evidence, heterogeneous structures, and insufficient generation grounding in multi-document question answering for fiscal regulations, this paper proposed a Multi-Document Heterogeneous Structural Graph-based Retrieval-Augmented Generation Method, called MDHSG-RAG. The method used structural templates to construct a unified hierarchical graph index. It performed coarse-to-fine hierarchical retrieval according to question types and used evidence chains to constrain answer generation. Experimental results showed that MDHSG-RAG outperformed the comparison methods in both retrieval and generation stages. The method improves evidence localization, answer accuracy, and traceability in fiscal regulation question answering.
文章引用：师培仁, 柴变芳. 基于多文档异构结构图的检索增强生成方法[J]. 计算机科学与应用, 2026, 16(9): 169-181. https://doi.org/10.12677/csa.2026.169298

参考文献

[1] 曹荣荣, 柳林, 于艳东, 等. 融合知识图谱的大语言模型研究综述[J]. 计算机应用研究, 2025, 42(8): 2255-2266.
[2] Lewis, P., Perez, E., Piktus, A., et al. (2020) Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks. Proceedings of the 34th International Conference on Neural Information Processing Systems, Vancouver, 6-12 December 2020, 9459-9474.
[3] Robertson, S. and Zaragoza, H. (2009) The Probabilistic Relevance Framework: BM25 and Beyond. Foundations and Trends® in Information Retrieval, 4, 1-174.
https://doi.org/10.1561/1500000019
[4] Karpukhin, V., Oguz, B., Min, S., Lewis, P., Wu, L., Edunov, S., et al. (2020) Dense Passage Retrieval for Open-Domain Question Answering. Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP), Online, 16-20 November 2020, 6769-6781.
https://doi.org/10.18653/v1/2020.emnlp-main.550
[5] Peng, B., Zhu, Y., Liu, Y., et al. (2024) Graph Retrieval-Augmented Generation: A Survey. ACM Transactions on Information Systems, 44, 1-52.
[6] Edge, D., Trinh, H., Cheng, N., et al. (2024) From Local to Global: A GraphRAG Approach to Query-Focused Summarization. arXiv:2404.16130.
[7] Guo, Z., Xia, L., Yu, Y., Ao, T. and Huang, C. (2025) LightRAG: Simple and Fast Retrieval-Augmented Generation. Findings of the Association for Computational Linguistics: EMNLP 2025, Suzhou, 4-9 November 2025, 10746-10761.
https://doi.org/10.18653/v1/2025.findings-emnlp.568
[8] Sarthi, P., Abdullah, S., Tuli, A., et al. (2024) RAPTOR: Recursive Abstractive Processing for Tree-Organized Retrieval. Proceedings of the 12th International Conference on Learning Representations, Vienna, 7-11 May 2024, 32628-32649.
[9] Wang, Y., Lipka, N., Rossi, R.A., Siu, A., Zhang, R. and Derr, T. (2024) Knowledge Graph Prompting for Multi-Document Question Answering. Proceedings of the AAAI Conference on Artificial Intelligence, 38, 19206-19214.
https://doi.org/10.1609/aaai.v38i17.29889
[10] Chen, W., Bai, T., Su, J., Luan, J., Liu, W. and Shi, C. (2025) Kg-Retriever: Efficient Knowledge Indexing for Retrieval-Augmented Large Language Models. 2025 IEEE International Conference on Knowledge Graph (ICKG), Limassol, 13-14 November 2025, 27-34.
https://doi.org/10.1109/ickg66886.2025.00011
[11] Gong, Z., Huang, Y. and Mai, C. (2025) MMRAG-DocQA: A Multi-Modal Retrieval-Augmented Generation Method for Document Question-Answering with Hierarchical Index and Multi-Granularity Retrieval. arXiv:2508.00579.
[12] Xiang, Z., Wu, C., Zhang, Q., et al. (2026) When to Use Graphs in RAG: A Comprehensive Analysis for Graph Retrieval-Augmented Generation. Proceedings of the 14th International Conference on Learning Representations, Brazil, 23-27 April 2026.
[13] 晋艳峰, 黄海来, 林沿铮, 等. 基于知识表示学习的KBQA答案推理重排序算法[J]. 计算机应用研究, 2024, 41(7): 1983-1991.
[14] Luo, L., Li, Y.F., Haffari, G. and Pan, S. (2024) Reasoning on Graphs: Faithful and Interpretable Large Language Model Reasoning. Proceedings of the 12th International Conference on Learning Representations, Vienna, 7-11 May 2024.
[15] Hu, Y., Lei, Z., Zhang, Z., Pan, B., Ling, C. and Zhao, L. (2025) GRAG: Graph Retrieval-Augmented Generation. Findings of the Association for Computational Linguistics: NAACL 2025, Albuquerque, 29 April-4 May 2025, 4145-4157.
https://doi.org/10.18653/v1/2025.findings-naacl.232
[16] He, X., Tian, Y., Sun, Y., Chawla, N., Laurent, T., Lecun, Y., et al. (2024) G-Retriever: Retrieval-Augmented Generation for Textual Graph Understanding and Question Answering. Advances in Neural Information Processing Systems 37, Vancouver, 10-15 December 2024, 132876-132907.
https://doi.org/10.52202/079017-4224
[17] Li, Z., Guo, Q., Shao, J., Song, L., Bian, J., Zhang, J., et al. (2025) Graph Neural Network Enhanced Retrieval for Question Answering of Large Language Models. Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers), Albuquerque, 29 April-4 May 2025, 6612-6633.
https://doi.org/10.18653/v1/2025.naacl-long.337
[18] Sarmah, B., Mehta, D., Hall, B., Rao, R., Patel, S. and Pasquali, S. (2024) Hybridrag: Integrating Knowledge Graphs and Vector Retrieval Augmented Generation for Efficient Information Extraction. Proceedings of the 5th ACM International Conference on AI in Finance, Brooklyn, 14-17 November 2024, 608-616.
https://doi.org/10.1145/3677052.3698671
[19] Li, Z., Deng, L., Liu, H., et al. (2024) UniOQA: A Unified Framework for Knowledge Graph Question Answering with Large Language Models. arXiv:2406.02110.
[20] Tang, X., Gao, Q., Li, J., et al. (2025) MBA-RAG: A Bandit Approach for Adaptive Retrieval-Augmented Generation through Question Complexity. Proceedings of the 31st International Conference on Computational Linguistics, Abu Dhabi, 19-24 January 2025, 3248-3254.