摘要
用户身份的唯一性标识是任何移动商务营销活动所必不可少的一项基础工作.由于多种因素的交叉影响,移动用户大进大出已成为移动运营商面临的普遍现象.一旦用户重新入网,不仅针对旧号码的用户特征刻画记录将完全荒废,而且面向新号码的用户洞察又需要从头开始,并将耗费相当长的时间才能得到完整的用户画像.而另一方面,基于用户资料的移动用户身份识别准确率仅为42%.因此,针对移动用户重入网身份识别的问题,提出基于用户网络访问项序的用户相似性计算方法,通过数据预处理、相似用户集裁剪、用户身份识别等6个步骤来精确定位重入网用户身份.在某电信运营商某地区25 809个用户60d网络访问日志这一数据集上,对所提方法进行了实验,总体准确率为98.32%,验证了方法的可行性与有效性.
User uniqueness identification is a kind of necessary work for any marketing activity in mobile commerce.However,as a result of many factors and their interactions,it is a common phenomenon for mobile operators that a great amount of mobile users enter communication network and exit.Once users re-enter the communication network,not only the descriptions of old users are totally useless,but also the work of inspecting new users has to be re-started and the whole user profile can only be obtained after a very long time.On the other hand,the accuracy rate of the user identification work based on user information is only 42%.Therefore,the mobile user identification problem is focused after they re-enter the communication network.And a similarity calculation method is presented based on users'network access item and procedure.The method can identify a user by 6 steps from pre-processing data → trimming the similar user set → identifying users,etc.. Experiments are made on a real data set of the network access log of 25 809 users from a communications corporation in 60 days.The overall accuracy rate of the presented identification method is 98.32%,which shows the feasibility and effectiveness of the method.
出处
《大连理工大学学报》
EI
CAS
CSCD
北大核心
2015年第2期223-228,共6页
Journal of Dalian University of Technology
基金
国家自然科学基金资助项目(71271037
71272093
71201014)
关键词
网络访问项序
移动用户
电信重入网
身份识别
用户相似性计算
network access item and procedure
mobile user
rejoining communication network
identity recognition
user similarity calculation