基于知识图谱的电力交易智能推荐技术研究  

Research on intelligent recommendation technology for power trading based on knowledge graph

作　　者：邓淑斌 王子石 梁志飞 牟春风 DENG Shubin;WANG Zishi;LIANG Zhifei;MOU Chunfeng(Guangzhou Electric Power Trading Center Co.,Ltd.,Guangzhou 510180,China)

机构地区：[1]广州电力交易中心有限责任公司,广东广州510180

出　　处：《粘接》2024年第10期149-152,共4页Adhesion

摘　　要：为提高电力交易智能推荐的准确性,提出一种基于知识图谱的电力交易智能推荐方法。构建了电力交易智能推荐知识图谱,并设计了基于互信息的抽取算法,将电力交易知识抽取分为概念对生成-排序和聚类压缩2步进行抽取,在Freebase知识库中对所提方法进行了验证。结果表明,所提方法可实现准确的知识抽取和答案抽取,抽取结果的准确率达到97.97%,平均精度均值为98%,平均质量得分为2.46。相较于随机抽取方法,抽取结果的平均精度均值提高了77.99%,平均质量得分提高了1.75,具有一定的有效性、准确性和优越性。To improve the accuracy of intelligent recommendation for power trading,a knowledge graph based intel⁃ligent recommendation method for power trading was proposed.The knowledge graph of intelligent recommendation of power trading was constructed,and the extraction algorithm based on mutual information was designed,and the knowledge extraction of power transaction was divided into two steps:concept generation-sorting and clustering com⁃pression,and the proposed method was verified in the Freebase knowledge base.The results showed that the pro⁃posed method could achieve accurate knowledge extraction and answer extraction,with an accuracy rate of 97.97%,an average accuracy of 98%,and an average quality score of 2.46.Compared to the random extraction method,the average accuracy of the extraction results had increased by 77.99%,and the average quality score had increased by 1.75,which has certain validity,accuracy and superiority.

关 键 词：知识图谱 电力交易 智能推荐 抽取算法 互信息 

分 类 号：TM7[电气工程—电力系统及自动化] TP39[自动化与计算机技术—计算机应用技术]