Multi-RAG:面向跨域数据的分布式检索增强生成框架

doi:10.11896/jsjkx.250900159

计算机科学 ›› 2026, Vol. 53 ›› Issue (6A): 250900159-7.doi: 10.11896/jsjkx.250900159

Multi-RAG:面向跨域数据的分布式检索增强生成框架

沈建伟, 陈汉林, 陈星

1 福州大学计算机与大数据学院福州 350116
2 福建省网络计算与智能信息处理重点实验室(福州大学) 福州 350116

出版日期:2026-06-16 发布日期:2026-06-12
通讯作者: 陈星(chenxing@fzu.edu.cn)
作者简介:(jwshenfzu@163.com)
基金资助:
国家自然科学基金(62072108);福建省促进海洋与渔业产业高质量发展专项资金(FJHYF-ZH-2023-02);福建省技术创新重点攻关及产业化项目 (2024XQ004);数据空间技术与系统全国重点实验室资助(QZQC2024007)

Multi-RAG:Distributed Retrieval-augmented Generation Framework for Cross-domain Data

SHEN Jianwei, CHEN Hanlin, CHEN Xing

1 College of Computer and Data Science,Fuzhou University,Fuzhou 350116,China
2 Fujian Key Laboratory of Network Computing and Intelligent Information Processing(Fuzhou University),Fuzhou 350116,China

Published:2026-06-16 Online:2026-06-12
About author:SHEN Jianwei,born in 2001,postgra-duate.His main research interests include large language models and know-ledge graphs.
CHEN Xing,born in 1985,Ph.D,professor,Ph.D supervisor,is a member of CCF(No.35725M).His main research interests include system software,software self-adaptation and cloud computing.
Supported by:
National Natural Science Foundation of China(62072108),Special Funds for Promoting High-quality Development of Marine and Fishery Industries in Fujian Province (FJHYF-ZH-2023-02),Fujian Key Technological Innovation and Industrialization Projects(2024XQ004) and National Key Laboratory of Data Space Technology and System(QZQC2024007).

摘要/Abstract

摘要： 随着大语言模型在自然语言处理任务中的深入应用,检索增强生成(RAG)技术已成为提升模型事实准确性的关键手段。然而,跨域数据的分散存储导致了数据汇聚障碍、规模扩展性不足以及分布式协同缺失等问题,使得传统集中式 RAG 框架难以满足跨域场景的需求。为此,提出了一种面向跨域数据的分布式检索增强生成框架——Multi-RAG。该框架允许各节点基于本地数据特性维护独立的嵌入模型与向量索引,并通过查询分发模块并行分发查询。各节点返回本地高分文档片段后,由全局重排序模块进行全局语义重排序,最终将筛选后的上下文输入大语言模型以生成答案。实验基于 MultiHop-RAG 数据集构建分布式环境展开,Multi-RAG 的 Hits@10 指标达到 0.765 9,较单一节点检索的 0.445 2 提升了 72%,与集中式方案的性能差异控制在 3.1% 以内;基于 DeepSeek-R1 模型的生成准确率较单节点方案提升了 48%。研究表明,该框架通过轻量级分布式协同机制与全局信息融合策略,在避免原始数据整合的同时,有效提升了跨域场景下的检索与生成性能,为解决跨机构、跨领域数据的协同知识利用提供了可行方案。

关键词: 分布式检索增强生成, 跨域数据协同, 全局重排序, 大语言模型, 信息融合

Abstract: The increasing application of large language models(LLMs) in natural language processing tasks has established retrieval-augmented generation(RAG) as a critical technique for enhancing factual accuracy.However,the distributed storage of cross-domain data presents significant challenges,including barriers to data aggregation,insufficient scalability,and the lack of native distributed coordination.These challenges render traditional centralized RAG frameworks inadequate for cross-domain scenarios.To address this limitation,a distributed RAG framework named Multi-RAG is proposed for handling cross-domain data.This framework enables individual nodes to maintain independent embedding models and vector indexes tailored to their local data characteristics.Queries are routed in parallel to relevant nodes through a query distribution module.Each node retrieves and returns locally high-scoring document segments.A global re-ranking module then performs global semantic reranking on these segments.The optimally re-ranked context is subsequently fed into the LLM for answer generation.Experiments conducted within a synthetically constructed distributed environment using the MultiHop-RAG dataset demonstrate Multi-RAG's effectiveness.The framework achieves a Hits@10 score of 0.765 9,representing a 72% improvement over single-node retrieval(0.445 2) and maintaining performance within 3.1% of a centralized approach.Answer generation accuracy using the DeepSeek-R1 model marks a 48% increase compared to the single-node baseline.The study indicates that through its streamlined distributed coordination mechanism and global information fusion strategy,Multi-RAG effectively enhances retrieval and generation performance in cross-domain settings without requiring raw data consolidation.This framework provides a practical and efficient solution for collaborative knowledge utilization across institutions and domains.

Key words: Distributed retrieval-augmented generation, Cross-domain data collaboration, Global re-ranking, Large language mo-del, Information fusion

中图分类号:

TP391

沈建伟, 陈汉林, 陈星. Multi-RAG:面向跨域数据的分布式检索增强生成框架[J]. 计算机科学, 2026, 53(6A): 250900159-7. https://doi.org/10.11896/jsjkx.250900159

SHEN Jianwei, CHEN Hanlin, CHEN Xing. Multi-RAG:Distributed Retrieval-augmented Generation Framework for Cross-domain Data[J]. Computer Science, 2026, 53(6A): 250900159-7. https://doi.org/10.11896/jsjkx.250900159

参考文献

[1] LEWIS P,PEREZ E,PIKTUS A,et al.Retrieval-augmentedgeneration for knowledge-intensive nlp tasks[J].Advances in Neural Information Processing Systems,2020,33:9459-9474.
[2] CUCONASU F,TRAPPOLINI G,SICILIANO F,et al.Thepower of noise:Redefining retrieval for rag systems[C]//Proceedings of the 47th International ACM SIGIR Conference on Research and Development in Information Retrieval.2024:719-729.
[3] ASAI A,WU Z,WANG Y,et al.Self-RAG:Learning to Retrieve,Generate,and Critique through Self-Reflection[C]//The Twelfth International Conference on Learning Representations.2024:1-30.
[4] ARSLAN M,GHANEM H,MUNAWARS,et al.A Survey on RAG with LLMs[J].Procedia Computer Science,2024,246:3781-3790.
[5] DU X Y,LI T,LU W,et al.Cross-domain data management[J].Computer Science,2024,51(1):4-12.
[6] LIU X,WANG R,SUN D,et al.Uncovering cross-domain re-commendation ability of large language models[C]//Companion Proceedings of the ACM on Web Conference 2025.2025:2736-2743.
[7] WAN Z,DING C,YU L,et al.A blockchain-based storage and sharing model for big health data[J].Journal of Chinese Mini-Micro Computer Systems,2023,44(3):636-645.
[8] ADDISON P,NGUYEN M T H,MEDAN T,et al.C-fedrag:A confidential federated retrieval-augmented generation system[J].arXiv:2412.13163,2024.
[9] SUBRAMANYA S J,DEVVRIT,SIMHADRI H V,et al.DiskANN:Fast Accurate Billion-point Nearest Neighbor Search on a Single Node[C]//Advances in Neural Information Processing Systems.2019:13748-13758.
[10] WANG S,KHRAMTSOVA E,ZHUANG S,et al.Feb4rag:Evaluating federated search in the context of retrieval augmented generation[C]//Proceedings of the 47th International ACM SIGIR Conference on Research and Development in Information Retrieval.2024:763-773.
[11] MALKOV Y A,YASHUNIND A.Efficient and robust approximate nearest neighbor search using hierarchical navigable small world graphs[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2018,42(4):824-836.
[12] JIN C,PENG H,ZHANG A,et al.Rankflow:A multi-role collaborative reranking workflow utilizing large language models[C]//Proceedings of the ACM on Web Conference 2025.2025:2484-2493.
[13] IZACARD G,GRAVE E.Leveraging Passage Retrieval with Generative Models for Open Domain Question Answering[C]//Proceedings of the 16th Conference of the European Chapter of the Association for Computational Linguistics.2021:874-880.
[14] HU E J,SHEN Y,WALLIS P,et al.Lora:Low-rank adaptation of large language models[J].ICLR,2022,1(2):3.
[15] GHAYVAT H,PANDYA S,BHATTACHARYA P,et al.CP-BDHCA:Blockchain-based Confidentiality-Privacy preserving Big Data scheme for healthcare clouds and applications[J].IEEE Journal of Biomedical and Health Informatics,2021,26(5):1937-1948.
[16] GOUNARI M,STERGIOPOULOS G,PIPYROS K,et al.Harmonizing open banking in the European Union:An analysis of PSD2 compliance and interrelation with cybersecurity frameworks and standards[J].International Cybersecurity Law Review,2024,5(1):79-120.
[17] LEE J S,JUN S P.Privacy-preserving data mining for open go-vernment data from heterogeneous sources[J].Government Information Quarterly,2021,38(1):101544.
[18] VOIGT P,VON DEM BUSSCHE A.The eu general data protection regulation(gdpr)[M].Cham:Springer International Publishing,2017.
[19] BONTA R.California consumer privacy act(CCPA)[EB/OL]https://oag.ca.gov/privacy/ccpa.
[20] Data Security Law of the People's Republic of China[J].Gazette of the Standing Committee of the National People's Congress of the People's Republic of China,2021(5):951-956.
[21] Personal Information Protection Law of the People's Republic of China[J].Gazette of the Standing Committee of the National People's Congress of the People's Republic of China,2021(6):1117-1125.
[22] AUMÜLLER M,BERNHARDSSON E,FAITHFULL A.ANN-Benchmarks:A benchmarking tool for approximate nearest neighbor algorithms[J].Information Systems,2020,87:101374.
[23] YANG Q,LIU Y,CHEN T,et al.Federated machine learning:Concept and applications[J].ACM Transactions on Intelligent Systems and Technology,2019,10(2):1-19.
[24] CALLAN J.Distributed information retrieval[M]//Advances in Information Retrieval:Recent Research from the Center for Intelligent Information Retrieval.Boston,MA:Springer US,2002:127-150.
[25] ASAI A,MIN S,ZHONG Z,et al.Retrieval-based languagemodels and applications[C]//Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics.2023:41-46.
[26] HUANG G.Data Internet:Digital Space Infrastructure[J].Communications of the China Computer Federation,2021,17(12):60-62.
[27] KAHN R,WILENSKY R.A framework for distributed digital object services[J].International Journal on Digital Libraries,2006,6(2):115-123.
[28] ZHANG N,LIU Y,MA X J,et al.Data Internet identification and resolution technology for human-machine-object integration[J].Journal of Software,2024,35(10):4681-4695.
[29] MCMAHAN B,MOORE E,RAMAGE D,et al.Communication-efficient learning of deep networks from decentralized data[C]//Artificial Intelligence and Statistics.PMLR,2017:1273-1282.
[30] WU W L,YIN H L,WANG N,et al.Cross-domain heterogeneous data query framework with collaboration of large language models and knowledge graphs[J].Journal of Computer Research and Development,2025,62(3):605-619.
[31] TANG Y,YANG Y.MultiHop-RAG:Benchmarking Retrieval-Augmented Generation for Multi-Hop Queries[C]//First Conference on Language Modeling.2024:1-16.

Metrics

Viewed

Full text

Abstract

Cited

Shared

Discussed

Multi-RAG:面向跨域数据的分布式检索增强生成框架

Multi-RAG:Distributed Retrieval-augmented Generation Framework for Cross-domain Data

PDF (PC)

摘要/Abstract

引用本文

使用本文

参考文献

相关文章 0

Metrics

本文评价

推荐阅读 0