大数据环境下精准客户定位的社交网络分析  

Precision customer positioning oriented social network analysis under big data environment

作　　者：曾小青[1,2] 张若欣 欧阳文光 王琪[1] ZENG Xiaoqing;ZHANG Roxin;OUYANG Wenguang;WANG Qi(School of Economics and Management, Changsha University of Science and Technology, Changsha 410004, China;Department of Mathematics and Computer Science, Northern Michigan University, MI 49855, USA;Qihu 360 Software Co. Ltd., Beijing 100015, China)

机构地区：[1]长沙理工大学经济与管理学院,长沙410004 [2]美国北密歇大学数学与计算机科学系,马凯特美国49855 [3]奇虎360科技有限公司,北京100015

出　　处：《计算机工程与应用》2017年第15期85-94,共10页Computer Engineering and Applications

基　　金：国家社会科学基金(No.13BGL063)

摘　　要：大数据为企业进行精准营销提供了重要支撑,精准营销能提升营销效果,提高客户满意度,精准营销的前提是客户识别与选择。通过分析网络个体与群体特征,社交网络分析能够定位核心价值客户。首先对社交网络的中心性进行分析,探讨社交网络节点地位与营销效果的关系,运用社群识别方法,对社交网络进行分群,提出并用Map Reduce实现了针对大规模社交网络的社群划分RMCL方法。在此基础上,构建了客户影响度与客户影响因子等指标,并结合中心度指标,定位社群的核心节点,并采用分类回归树方法,研究了社交网络结构与客户消费响应关系,并确定了变量重要性,为企业采取客户差异化营销组合策略提供指导。Big data provides important support for enterprises doing precision marketing decision.It can significantlyenhance marketing effectiveness and improve customer satisfaction.Customer identification and selection is fundamentalfor precision marketing.It helps companies target valuable customers and design marketing mix strategies.As a socialstructure analysis method,Social Network Analysis(SNA)method can effectively target the key customers by analyzingcustomer’s individual characteristics and group features.Firstly,through the social network centrality analysis,this paperstudies the relationship between the node centrality in a network and the marketing effectiveness.Furthermore,by introducinga method called RMCL(Regularized Markov Clustering),a solution is presented to partition social network intocommunities.More importantly,in order to detect communities in large-scale social networks,the RMCL method is implementedin Map Reduce Pattern,a parallel programming framework.This paper also explores the approaches to identifythe core node within a social network community.Two indicators are proposed in term of customer influence degree andinfluence factor to measure the customer’s impact on purchasing.Lastly,combining with social network centralitymetrics,CART method is used to uncover the links between the social network and customer consuming response,andthe order of variable importance is determined,which provides a feasible guidance for the enterprises adopting differentiatedmarketing strategies.

关 键 词：社交网络分析 精准营销 客户定位 大数据 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]