基于深层标签和K-Means++算法的电力用户画像研究

Research on Power User Portrait Based on Deep Labeling and K-Means++Algorithm

以电力用户的基础属性、电力公司业务办理系统数据以及95598工单数据为基础数据源,经过数据预处理后,采用自然语言处理的方法对用户咨询内容进行标签提取,建立用户画像标签库,进而提出基于深层标签和K-Means++算法的电力用户画像方法.实验结果表明,相比现有相关研究方法,构建的用户标签覆盖率较高、用户画像细粒度较高,F1值高达95.7%,更适合电力营业厅差异化服务客户.

The basic attributes of electric power users,the data of the electric power company’s business processing system and 95598 work order data are used as the basic data sources.After data pre-processing,a natural language processing method is used to extract labels from the users’consultation to establish a user portrait label library.Then an improved LSTM-based K-Means clustering algorithm is proposed to realize the construction of electric power user portraits based on deep labels and the K-Means++algorithm for electricity user portraits.The experimental results show that,compared with the existing related research results,the user label coverage rate and user portrait fine-grained constructed by the method of this paper is high,and the F1-Score is as high as 95.7%,which is more suitable for the differentiated service customers of the electric power business hall.