基于无监督学习的学生用户群体发现  

Student user group discovery based on unsupervised learning

作　　者：欧阳秀平[1] 林敏[1] 陈祥[1] 叶浩然 江昊[2] OUYANG Xiuping;LIN Min;CHEN Xiang;YE Haoran;JIANG Hao(China Unicom Guangdong Branch,Guangzhou 510627,China;School of Electronic Information,Wuhan University,Wuhan 430074,China)

机构地区：[1]中国联合网络通信有限公司广东省分公司,广东广州510627 [2]武汉大学电子信息学院,湖北武汉430074

出　　处：《武汉大学学报（工学版）》2024年第11期1652-1658,共7页Engineering Journal of Wuhan University

摘　　要：结合学生用户群体的行为特征和运营商记录的用户数据,基于无监督社群识别方法实现了对学生用户的群体发现。将学生用户群体分为学龄前儿童、中小学生、高校学生3类。学龄前儿童和中小学生根据教育相关应用程序的使用数据提取对应特征,经过聚类分析,将符合预期的用户标记为学生;高校学生根据用户在高校范围内的活动情况提取特征,经过聚类分析,将符合预期的用户标记为学生。该方法可以识别各年龄段的学生用户和家长用户,辅助进行学生或儿童的相关产品推荐或市场分析。According to the behavior characteristics of student user groups and user data recorded by operators,this paper realizes group discovery of student users based on unsupervised community identification method.The student user groups are divided into three categories:preschool children,primary and secondary school students,and college students.The corresponding characteristics of preschool children and primary and secondary school students are extracted according to the service data of education related applications,and the users who meet the expectations are marked as students through cluster analysis.According to the activities of users in the scope of colleges and universities,the characteristics of college students are extracted,and the users who meet expectations are labeled as students through cluster analysis.This method can identify student users and parent users of all ages to assist in student/child related product recommendation or market analysis.

关 键 词：无监督学习 聚类 学生用户 群体发现 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程]