海量知识图谱分布式管理与查询技术  

Distributed Management and Query Technology for Massive Knowledge Graph

作　　者：李彭伟[1] LI Pengwei(The 28th Research Institute of China Electronics Technology Group Corporation,Nanjing 210007,China)

机构地区：[1]中国电子科技集团公司第二十八研究所,南京210007

出　　处：《指挥信息系统与技术》2021年第2期75-80,93,共7页Command Information System and Technology

基　　金：装备发展部“十三五”预研课题资助项目。

摘　　要：随着知识图谱的规模日益扩大,存储难和计算效率低成为制约知识图谱发展的重要因素。基于图谱结构特点,提出了基于密度的知识图谱节点聚类算法,以及适用于知识图谱结构的节点标记和处理方法。针对星形查询,设计了跨集群管理方法,并构建了知识图谱分割和子集群间索引。试验表明,该算法可实现海量知识图谱的分布式管理与查询。With the expansion of the scale of the knowledge graph,the difficult storage and the low computing efficiency have become the important factors restricting the development of knowledge graph.Based on the features of knowledge graph,the density-based knowledge graph node clustering algorithm is proposed,and a method of node marking and processing for knowledge graph structure is put forward.For star query,a cross-cluster management method is designed,and the knowledge graph segmentation and the inter-subset index are constructed.Experiments show that the algorithm can realize the distributed management and query of massive knowledge graph.

关 键 词：知识图谱 存储 图分割 

分 类 号：TP123[自动化与计算机技术—控制理论与控制工程]