移动互联环境下基于改进模糊聚类理论的用户分群算法  

User clustering algorithm based on improved fuzzy clustering theory in mobile Internet environment

作　　者：张晖[1,2] 征原 Zhang Hui;Zheng Yuan(Jiangsu Provincial Key Laboratory of Wireless Communications,Nanjing University of Posts&Telecommunications,Nanjing 210003,China;Jiangsu Provincial Key Laboratory of Computer Information Processing Technology,Soochow University,Suzhou Jiangsu 215006,China)

机构地区：[1]南京邮电大学江苏省无线通信重点实验室,南京210003 [2]苏州大学江苏省计算机信息处理技术重点实验室,江苏苏州215006

出　　处：《计算机应用研究》2018年第10期3033-3036,共4页Application Research of Computers

基　　金：国家自然科学基金资助项目(61471203);江苏省"青蓝工程"资助项目(2016);南京邮电大学"1311"人才计划资助项目(2015);江苏省计算机信息处理技术重点实验室开放课题(KJS1518);国家科技重大专项资助项目(2012ZX03001008-003)

摘　　要：在移动互联环境下,依据用户行为规律对业务兴趣相似用户进行分群,可为业务准确推荐和资源有效配置提供有力支撑。因此,提出一种基于改进模糊聚类理论的用户分群算法。首先,分别定义业务兴趣相似度和业务顺序相似度,进而建立用户综合相似度指标;其次,构建基于用户综合相似度的模糊聚类模型,进而采用网格划分方法确定初始群组中心,并依据平均用户隶属度调整用户群组个数,从而实现快速准确的用户分群。仿真结果验证了该算法的有效性。In mobile Internet environment,users with similar service interests are grouped according to their behavior law,which provides the strong support for accurate service recommendation and effective resource allocation.Therefore,this paper proposed an improved fuzzy clustering theory based user clustering algorithm.First,it respectively defined service interest si-milarity and service sequence similarity,and then established integrated similarity metric.Second,it formulated a integrated similarity metric based fuzzy clustering model,which determined the initial cluster centers by using the grid partition method and adjusted the user cluster number according to the average user membership,so as to obtain user clusters rapidly and accurately.Simulation results verify the effectiveness of proposed algorithm.

关 键 词：移动互联环境 行为分析 用户分群 模糊聚类 

分 类 号：TN915[电子电信—通信与信息系统]