Research On User Identity Linkage Algorithms Among Social Networks

With the rapid development of online social networking and the advent of the era of big data,data mining technology is more and more widely applied in social network research.The diversity of social network functions and the diversity of user needs make users active in multiple social networks at the same time.Data mining in a single social network can‘t show the global characteristics of the user in the whole social network.Only by linking accounts in different social networks with their corresponding real user identity,researchers can carry out comprehensive and complete user information mining in the multi-layer network,and thus they can obtain the global characteristics of the users in the network.This technology to link accounts in multiple networks to their corresponding user identities is called user identity linkage technology among social networks.The study of user identity linkage technology among social networks can make researchers get the overall view of social network,provide the possibility for further research of social network,and have great help to maintain the Internet security environment.Because of the inconsistency of the user's profile information and the difference between the actual networks,it is difficult to link the accounts of different networks precisely by using the traditional attribute similarity calculation method only in the study of user identity linkage problem among social networks.In order to solve this problem,by analyzing the characteristics of social network and combining the profile information and structure information in the network,a Structure-based Neighbor Iterative Similarity Algorithm is proposed to solve the shortage of the traditional attribute similarity calculation method,and a more complete user identity linkage algorithm among social networks is obtained.The algorithm is a semi-supervised propagation method,which takes the priori seed links as the anchors and iteratively propagate the whole network to complete the work of the user's identity link.Using the data set of real social network and comparing with other algorithms,the performance of the algorithm is analyzed and discussed in this paper,and the excellent comprehensive performance of the algorithm is proved.In the later work,this paper optimizes the algorithm,and reduces the errors in the iteration through the threshold setting and the result cutting to improve the accuracy of the algorithm.The feasibility and effectiveness of the optimization method are verified by comparing the algorithms of pre optimization,partial optimization and complete optimization,and the specific effects of threshold,cutting ratio and priori seed links on the performance of the algorithm are discussed.