计算机科学

计算机科学 ›› 2026, Vol. 53 ›› Issue (5): 119-128.doi: 10.11896/jsjkx.250600019

• 数据库 & 大数据 & 数据科学 • 上一篇    下一篇

基于企业数据空间的数据资源组织方法与数据资产管理

李敏波1, 王少华2, 吴大臻1   

  1. 1 复旦大学计算与智能创新学院 上海 200433
    2 浪潮云洲工业互联网有限公司 济南 250101
  • 收稿日期:2025-05-29 修回日期:2025-08-11 发布日期:2026-05-08
  • 通讯作者: 王少华(wangshaohua03@inspur.com)
  • 作者简介:(limb@fudan.edu.cn)
  • 基金资助:
    国家重点研发计划(2023YFC3304400)

Data Resource Organization Method Based on Enterprise Dataspace and Data Asset Management

LI Minbo1, WANG Shaohua2, WU Dazhen1   

  1. 1 College of Computer Science and Artificial Intelligence, Fudan University, Shanghai 200433, China
    2 Inspur Yunzhou Industrial Internet Co., Ltd., Jinan 250101, China
  • Received:2025-05-29 Revised:2025-08-11 Online:2026-05-08
  • About author:LI Minbo,born in 1970,Ph.D,associate professor,is a member of CCF(No.53905M).His main research interests include industrial big data and industrial AI.
    WANG Shaohua,born in 1984,senior engineer.His main research interests include industrial Internet,AI and supply chain collaboration and data ser-vices.
  • Supported by:
    National Key R & D Program of China(2023YFC3304400).

PDF (PC)

237

摘要： 针对军工科研单位因保密分级分块管理导致的数据资源孤岛问题,以及数据资源智能检索与知识复用困难,提出了多源异构企业数据资源与数据资产的治理方案。通过企业数据空间属性图模型实现数据资源之间的图节点关联映射,构建融合BOM树型结构的数据资源知识图谱,涵盖研发工艺、生产制造与质检数据的层级关系、属性信息及关联关系。具体地,提出了一个新颖的RAG框架HireRAG,建立了基于C-HNSW的知识图谱社群化分层索引,低层保留细粒度的知识单元,高层社群提供全局摘要,以处理不同层次的检索;提出了一种图增强聚类算法,使得C-HNSW更好地捕捉知识图谱中的语义信息。实验表明,HireRAG相比现有的一些先进RAG框架更适合处理企业数据空间BOM关联数据,可以在实现最高准确率的同时保证检索效率优势。数据资产管理系统确保数据资产全过程合规入表。

关键词: 企业数据空间, 数据资源, 知识图谱, 数据检索, HNSW

Abstract: Aiming at the problem of data resource islands caused by confidentiality hierarchical and block management in military research institute,and the difficulty of intelligent retrieval and knowledge reuse of data resources,a governance solution for multi-source heterogeneous enterprise data resources and data assets is proposed.The graph node association mapping between data resources is realized through the attribute graph model of enterprise data space,and a data resource knowledge graph integrating the BOM tree structure is constructed,covering the hierarchical relationship,attribute information and association relationship of R&D process,production and manufacturing,and quality inspection data.This paper proposes a novel RAG framework－HireRAG,and establishes a community-based hierarchical index of knowledge graph based on C-HNSW.The low-level retains fine-grained knowledge units,and the high-level community provides a global summary to handle retrieval at different levels.A graph-enhanced clustering algorithm is proposed to enable C-HNSW to better capture the semantic information in the know-ledge graph.Experiments demonstrate that HireRAG is more adapt at processing bill of materials(BOM) related data within enterprise data spaces compared to several existing advanced retrieval-augmented generation(RAG) frameworks.Furthermore,it achieves superior performance metrics in both retrieval recall and accuracy.The data asset management system ensures that the data assets are entered into the table in compliance with the whole process.

Key words: Enterprise dataspace, Data resources, Knowledge graph, Data query, Hierarchical Navigable Small World Graphs

中图分类号: 

  • TP391
[1]FRANKLIN M,HALEVY A,MAIER D.From databased todataspaces:A new abstraction for information management[J].SIGMOD Record,2005,34(4):27-33.
[2]FAN S H,HOU M S.Dataspace:A New Data Organization and Management Model[J].Computer Science,2023,50(5):115-127.
[3]DITTRICH J P,SALLES M A V.iDM:A unified and versatile data model for personal dataspace management[C]//Internatio-nal Conference on Very Large Data Bases(VLDB’06).2006:367-378.
[4]WEN B L,JIAO S J,GUO J.Research on Data OrganizationMethod Based on Enterprise DataSpace[J].Computer Technology and Development,2020,30(12):56-60.
[5]LI Y K,MENG X F,ZHANG X Y.Research on data space technology[J].Journal of Software,2008,19(8):2018-2031.
[6]YAN X F,YU P S,HAN J W.Graph Indexing:A frequent structure-based approach[C]//Proceedings of the 24th International Conference on Management of Data(SIGMOD 2004).New York:ACM,2004:335-346.
[7]HE H,SINGH A K.Closure-Tree:An index structure for graph queries[C]//Proceedings of the 22nd International Conference on Data Engineering(ICDE 2006).Dallas:IEEE Computer So-ciety,2006.
[8]HOLDER L,COOK D,DJOKO S.Substructure discovery in the subdue system[C]//Proceedings of the AAAI Workshop of Conference on Knowledge Discovery in Databases.Menlo Park:AAAI,1994:169-180.
[9]JIANG H L,WANG H X,YU P S,et al.GString:A novel approach for efficient search in graph databases[C]//Proceedings of the 23rd International Conference on Data Engineering(ICDE 2007).Dallas:IEEE Computer Society,2007:566-575.
[10]VAN BRUGGEN R,BATON J.Learning Neo4j 3.x-second edition[M]//Birmingham:Packt Publishing,2017:33-36.
[11]ELSAYED I,MUSLIMOVIC A,BREZANY P.Intelligent data spaces for science C WSEAS[C]//International Conference on Computational Intelligence,Man-Machine Systems and Cybernetics.2008:94-100.
[12]YANG D,SHEN D,NIE T,et al.Layered graph data model for data management of dataspace support platform[C]//International Conference on Web-Age Information Management.Berlin:Springer,2011:353-365.
[13]FRANKLIN M,HALEVY A,MAIER D.A first tutorial ondataspaces[C]//Proceedings of the VLDB Endowment.2008:1516-1517.
[14]XUE Q,ZENG X Q,OUYANG Z Y.Discussion on Key Issues of Data Asset Inclusion the Background of Digitalization and Intelligence[J].The Chinese Certified Public Accountant,2024,10:105-113.
[15]EDGE D,TRINH H,CHENG N,et al.From Local to Global:A Graph RAG Approach to Query-Focused Summarization[J].arXiv:2404.16130,2024.
[16]GUO Z R,XIA L H,YU Y H,et al.Lightrag:Simple and fast retrieval-augmented generation[J].arXiv:2410.05779,2024.
[17]GROVER A,LESKOVEC J.node2vec:Scalable feature learning for networks[C]//Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mi-ning.2016:855-864.
[18]XU X,YURUK N,FENG Z,et al.Scan:A structural clustering algorithm for networks[C]//Proceedings of the 13th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining.2007:824-833.
[19]WANG C,WANG F,ONEGA T.Network optimization ap-proach to delineating health care service areas:Spatially constrained Louvain and Leiden algorithms[J].Transactions in GIS,2021,25(2):1065-1081.
[20]KOJIMA T,GU S S,REID M,et al.Large language models are zero-shot reasoners[J].Advances in Neural Information Processing Systems,2022,35:22199-22213.
[21]ROBERTSON S E,WALKER S.Some simple effective approximations to the 2-poisson model for probabilistic weighted retrieval[C]//Proceedings of the Seventeenth Annual International ACM-SIGIR Conference on Research and Development in Information Retrieval(SIGIR’94).London:Springer,1994:232-241.
[22]SARTHI P,ABDULLAH S,TULI A,et al.Raptor:Recursiveabstractive processing for tree-organized retrieval[C]//The Twelfth International Conference on Learning Representations.2024.
[23]NUSSBAUM Z,MORRIS J X,DUDERSTADT B,et al.Nomic embed:Training a reproducible long context text embedder[J].arXiv:2402.01613,2024.
Viewed
Full text


Abstract

Cited
  Shared   
  Discussed   
No Suggested Reading articles found!
渝ICP备16013121号-4
地址：重庆市两江新区洪湖西路18号    邮编：401121  
电话：023-63500828（编辑部） 023-67039612（会议/专题） 023-67039623（财务/发票）
E-mail：jsjkx12@163.com
本系统由北京玛格泰克科技发展有限公司设计开发