摘要
为减少运营商用户投诉,实现对用户投诉的事前控制,可以使用决策树算法预测用户投诉的风险度及可能投诉的风险点。通过对大数据平台中海量的管道数据以及用户接触语音、上网行为、投诉行为等数据的分析,发现投诉用户的特征值。整合投诉用户的交际圈、忍耐度、情感度、业务表象等特征,最终形成用户投诉预测复合模型,预测用户整体投诉的风险度及可能投诉的风险点。借助触点平台,将模型滚动预判信息推送到客服坐席、营业前台、代理商等各服务渠道,当用户通过上述渠道进行咨询时,接待人员可以根据平台推送的投诉预警值、用户敏感内容、容忍度、历史接触记录等信息,执行相应措施,以有效减少普通投诉和升级投诉的产生。
In order to reduce the operator user complaints and realize the prior control of customer complaints,the decision tree algorithm is used to predict the risk degree of user complaints and the risk points of possible complaints.Through the analysis of the massive pipeline data in the big data platform,as well as the user contact voice,Internet behavior,complaint behavior and other data,the complaint user characteristic value of operators is found.Integrate the analysis of users’social circle,patience,emotion degree,business appearance and other characteristics of complaints,and finally form a composite model of user complaint prediction to predict the risk degree of users’overall complaints and the risk points of possible complaints.With the help of the contact platform,the model rolling prediction information is pushed to the customer service seats,business front desk,agents and other service channels.When users consult through the above channels,the reception staff can implement the corresponding measures according to the complaint warning value,user sensitive content,tolerance,historical contact records and other information pushed by the platform,to effectively reduce the generation of ordinary complaints and upgrade complaints.
作者
俞涛
郝洁
张小晖
YU Tao;HAO Jie;ZHANG Xiaohui(China United Network Communications Co.,Ltd.,Shijiazhuang 050011,China)
出处
《移动信息》
2023年第2期208-210,共3页
MOBILE INFORMATION
关键词
数理统计学
决策树算法
投诉用户延伸特征值
Mathematical statistics
Decision tree algorithm
Complaint user extension feature value
