摘要 针对电网企业的数据资源无法被智能分析与管理等问题,提出基于知识图谱的语义搜索(KGSS)算法。KGSS算法通过MPP采集模型和Madoop数据库对结构化、半结构化和非结构化数据进行知识抽取,构建知识实体、属性以及实体间关系。利用嵌入投影模型(PEM)建立多结构数据的电网知识库,实现支持语义搜索的知识图谱。KGSS算法采用相似性策略进行实施语义搜索。性能分析表明,相比于传统的关键词搜索,KGSS算法的查准率和召回率得到有效的提升。 To solve the problem that the data resources cannot be intelligently analyzed and managed,a knowledge graph-based semantic search(KGSS)algorithm is proposed in this paper.The KGSS algorithm firstly extracts structured,semi-structured and unstructured data through MPP acquisition model and Madoop database,and constructs knowledge entities,attributes and inter-entity relations.Projection embedding model(PEM)is used to build the knowledge base of grid with multi-structure data so as to realize the knowledge map supporting semantic search.Finally,semantic search is implemented based on similarity strategy.The performance analysis shows that compared with traditional keyword search,the precision and recall rate of KGSS algorithm are improved effectively.
机构地区 国家电网北京市电力公司
出处 《实验室研究与探索》 CAS 北大核心 2021年第4期71-74,86,共5页 Research and Exploration In Laboratory
关键词 智能电网 知识图谱 语义搜索 嵌入投影模型 相似性策略 smart grid security knowledge graph semantic search projection embedding model similarity strategy
