基于网络店铺订单数据的群体用户画像构建研究

A STUDY OF GROUP USER PORTRAIT CONSTRUCTION BASED ON ORDER DATA FROM ONLINE STORES

针对电子商务平台中网络店铺的订单数据,提出了一种用于精准客户细分的群体用户画像构建方法。该方法主要分为4个步骤:第一步,通过用户基本属性及领域词汇表创建用户画像模板;第二步,根据订单数据构建单用户画像实例,形成用户画像数据库;第三步,对单用户画像的各种类型属性指标进行规范化处理;第四步,通过属性相似度计算,运用聚类算法实现群体用户画像聚类。以淘宝网某"线上礼裙"企业店铺为例进行实验,结果表明:7个属性平均精确度为0.913,综合精确度为0.522,说明该方法是有效的。

Based on order data of online stores on E-Commerce platform, this paper proposes the method of group user portrait for more precise customer segmentation. This method is divided into four steps:(1) creating portrait templates according to user profile and domain vocabulary;(2) constructing single user portrait as an example based on order data so as to form a database;(3) standardizing all kinds of attribute indices of single user;(4) clustering group users by calculating attribute similarity via clustering algorithm. The paper makes a case study of an online dress store from Taobao platform, and draws that the average precision of the seven attributes is 0.913 and that the comprehensive precision is 0.522, indicating effectiveness of the method of group user portrait.