基于社交网络分析的流量红包客户挖掘与传播模式  

Data red envelope clients mining and communication model based on social network analysis

作　　者：徐海勇 陶涛 黄岩 唐崔巍 张兆静 吴晶 XU Haiyong;TAO Tao;HUANG Yan;TANG Cuiwei;ZHANG Zhaojing;WU Jing(China Mobile Information Technology Co.,Ltd.,Beijing 100037,China)

机构地区：[1]中移动信息技术有限公司,北京100037 [2]中移动信息技术有限公司电子渠道运营中心,北京100037

出　　处：《电信科学》2020年第8期139-150,共12页Telecommunications Science

摘　　要：电信运营商持续创新流量经营模式与手段,将流量营销与社交化的红包活动相结合,创新推出流量红包活动,激发客户流量使用兴趣。面对社交化流量红包客户特征与传播模式的研究痛点及当前社群建模算法较单一的技术现状,详细研究、对比6种社群建模算法的应用效果,筛选出适合流量红包的最优算法,并定位分析核心价值客户群的特征。仿真结果显示,Multi-Level算法在流量红包场景中表现更好,基于该算法挖掘种子客户、高价值客户、低价值客户和沉默客户4种特征客户群的社交网络结构。社交网络分群结论为运营商精准营销、精准推送等营销活动以及沉默客户促活、流失客户挽回等客户运营管理提供了有效指导。Telecom operators innovate continually in data management modes and combine data marketing with social red envelope activities,consequently launching data red envelope activities,which stimulates customers’interest of data using.Faced with the research pain points of customer characteristics and propagation modes of socialized traffic red envelopes and the current status of the single community modeling algorithm,the detailed research and comparison of the application effects of the six community modeling algorithms were conducted to select the optimal traffic envelope algorithm,and the characteristics of the core value customer groups were located and analyzed.Results show that Multi-Level algorithm performs better in data red-envelope scenario.Based on this algorithm,four characteristic customer groups were mined,namely seed customer,high-value customer,low-value customer and silent customer.The social network clustering conclusion provided effective guidance for operators’precise marketing,precise push and other marketing activities,as well as other customer operation management such as silent customer activation,lost customer recovery and so on.

关 键 词：社交网络分析 流量红包 客户群划分 

分 类 号：TP301[自动化与计算机技术—计算机系统结构]