LBSN协作式个性化链接预测算法  

Cooperation based personalized link prediction algorithm in LBSN

作　　者：胡敏[1,2] 崔永胜 黄宏程 陈元会[1] Hu Min;Cui Yongsheng;Huang Hongcheng;Chen Yuanhui(School of Communication&Information Engineering,Chongqing University of Posts&Telecommunications,Chongqing 400065,China;Chongqing Engineering Research Center of Communication Software,Chongqing 400065,China)

机构地区：[1]重庆邮电大学通信与信息工程学院,重庆400065 [2]重庆市通信软件工程技术研究中心,重庆400065

出　　处：《计算机应用研究》2020年第4期1188-1193,共6页Application Research of Computers

基　　金：国家自然科学基金资助项目(61871062);重庆邮电大学科研基金资助项目(A2018-07)。

摘　　要：在基于位置的社交网络中用户链接与位置链接之间具有一定的内在关联,而且不同的用户在社交网络中的表现也存在差异,因此对于以上问题提出一种协作式个性化链接预测算法。针对用户的个性化特征,采用核密度估计方式对用户在时间和空间维度建模,基于兴趣组对用户进行重叠社团划分,并通过社团、好友以及签到关系进行个性化用户链接预测;基于个性化用户链接预测结果,利用从社团重启的随机游走预测用户的个性化位置链接;协作式个性化链接预测算法通过用户链接预测和位置链接预测的迭代使得两者性能相互提升。实验结果表明,所提算法相比于现有算法具有更好的预测性能。There is a certain internal relationship between user links and location links in location-based social network(LBSN),and different users also have different behaviors in the network.Therefore,view of the above problem,this paper proposed a cooperation based personalized link prediction algorithm(CPP)in LBSN.For the user’s personalized features,the algorithm used the kernel density estimation method to model the user’s time and spatial dimensions.It used the interest groups to divide the users into overlapping communities,and performed the personalized user link prediction through the community,friends and sign-in relationships.Based on the prediction of the personalized user link,it predicted personalized link relationship between users and locations via the random walk algorithm with community restarting.The CPP algorithm improved the performance by the iteration of the user link prediction and the location link prediction.The experimental results show that the CPP algorithm has better prediction performance than that of the existing algorithms.

关 键 词：链接预测 基于位置的社交网络 核密度估计 个性化 随机游走 

分 类 号：TN915.03[电子电信—通信与信息系统]