基于TF-IDF算法的运营商用户画像分析  

Carrier user portrait analysis based on TF-IDF algorithm

作　　者：刘佳 陈敏时 谢懿 苏昭玉 武星宇 LIU Jia;CHEN Min-shi;XIE Yi;SU Zhao-yu;WU Xing-yu(China Mobile Group Co.,Ltd.,Beijing 100032,China;China Mobile Group Design Institute Co.,Ltd.,Beijing 100080,China)

机构地区：[1]中国移动通信集团有限公司,北京100032 [2]中国移动通信集团设计院有限公司,北京100080

出　　处：《电信工程技术与标准化》2023年第10期1-5,30,共6页Telecom Engineering Technics and Standardization

摘　　要：随着通信技术的迅速发展,海量日志的产生给企业运营带来了安全隐患,除了能否及时检测出电信运营商企业内各类业务系统及设备的海量日志是否安全、合规之外,基于用户行为的分析对日志安全也尤为重要。为了直观掌握用户行为,本文基于TF-IDF标签权重算法对运营商海量用户业务日志数据构建用户画像。根据实际业务场景构建RFS模型,通过开发的用户画像标签体系分析研究运营商用户行为,对用户操作日志中的潜在高风险信息进行数据挖掘,并输出可视化用户行为分析画像结果。实际应用表明,通过用户画像能识别风险用户行为信息,同时也能为日志安全分析提供参考价值,帮助企业实现数据安全运营动态监测。With the rapid development of communication technology,the generation of massive logs has brought security risks to enterprise operations.In addition to whether the massive logs of various business systems and devices in telecom operators can be timely detected whether they are safe and compliant,the analysis based on user behavior is also particularly important for log security.In order to intuitively grasp the user behavior,the TFIDF label weighting algorithm is used to construct user portraits from the massive user business log data of operators.According to the actual business scenario,the RFS model is constructed to develop the user portrait label system.Data mining of potentially high-risk information in user operation logs and output visual user behavior analysis profi le results.The practical application shows that the user portrait modeling can mine the behavior information of users at risk,and also can provide reference value for log security audit,and help enterprises realize the dynamic monitoring of data security operation.

关 键 词：日志安全 用户行为分析 用户画像 标签权重算法 数据挖掘 

分 类 号：TN929.5[电子电信—通信与信息系统]