基于知识图谱与优化数据架构的电力服务系统设计

Design of power service system based on knowledge atlas and optimized data architecture

为了提升电力服务的智能化水平,文中基于软件工程理论完成了线上电力服务系统设计,主要包括在线服务接入、人工交互管理、知识图谱管理及系统支撑管理共4个模块。针对知识图谱中噪声数据多且较为稀疏的特点,将知识图谱的查询问题转化为图论中的子图匹配问题。通过引入邻域向量的概念,以相似度作为查询依据来有效提高系统的效率。相较于RDF-3X与NESS算法,所提算法在查询耗时上分别降低了85.59%和73.71%。此外,基于FastDFS分布式存储架构实现了知识图谱的分布式存储。且系统数据库的SQL查询、知识图谱的响应时间与单节点数据架构相比分别降低了58.3%和91.7%,故其具备良好的扩展性,并可满足未来数据增长需求。

In order to improve the intelligent level of power service,this paper completes the design of online power service system based on software engineering theory.The system includes four modules:online service access,manual interaction management,knowledge map management and system support management.In view of the characteristics of more noise data and sparse data in the knowledge map,this paper transforms the query problem of the knowledge map into the sub graph matching problem in graph theory,introduces the concept of neighborhood vector,and takes the similarity as the query basis,which effectively improves the system efficiency.Compared with RDF-3X and NESS algorithms,the query time is reduced by 85.59%and 73.71%respectively.In addition,the distributed storage of knowledge map is realized based on FastDFS distributed storage architecture.The response time of SQL query and knowledge map of the system database is reduced by 58.3%and 91.7%respectively compared with the single node data architecture,and has good scalability,which can meet the needs of data growth in the future.