融合情境因素的电子商务用户兴趣挖掘仿真  

E-Commerce User Interest Mining Simulation Based on Contextual Factors

作　　者：成保梅[1] 韩景灵[1] CHENG Bao-mei;HAN Jing-ling(Business College,Shanxi University,Shanxi Taiyuan 030031,China)

机构地区：[1]山西大学商务学院,山西太原030031

出　　处：《计算机仿真》2020年第4期326-329,共4页Computer Simulation

基　　金：院级科研资金资助项目(编号2018016):融合情境因素的电子商务推荐方法研究。

摘　　要：针对当前电子商务用户兴趣挖掘过程,普遍存在准确度较低、综合性能较差以及召回率较低等问题。提出融合情境因素下基于网络结构的电子商务用户兴趣挖掘方法。通过计算融合情境相似度来获取电子商务用户当前情境的近似情境集,对电子商务用户-兴趣项-情境构建三维模型,采用情境预过滤方法对近似情境集进行降维处理,以此得到每种主题的兴趣度。根据兴趣度大小将电子商务用户模式划分为低兴趣度、一般兴趣度、高兴趣度三个级别,保留电子商务用户兴趣度较高的模式,删除较低的模式,完成电子商务用户兴趣挖掘。实验结果表明,所提出方法用户兴趣挖掘准确率较高、综合性能较好、召回率较高。Due to low accuracy, poor comprehensive performance and low recall rate of current method, this paper proposes a method to mine e-commerce user interest with situational factors based on network structure. By calculating the similarity of fusion context, we obtained the approximate situation set of e-commerce users in current situation. Then, we constructed the three-dimensional model of e-commerce users-interest items-context. Meanwhile, we used situational pre-filtering to reduce the dimension of approximate situation set, so as to get the interest degree of each topic. According to the interest degree, we divided the e-commerce user model into three levels:low interest, general interest and high interest degree. Finally, we retained the mode with high interest degree and deleted the mode with low interest degree. Thus, we completed the e-commerce user interest mining. Simulation results show that the proposed method has higher accuracy, better comprehensive performance and higher recall rate during the user interest mining.

关 键 词：融合情境因素 电子商务 用户 兴趣挖掘 

分 类 号：TP393.09[自动化与计算机技术—计算机应用技术]