Person Identification between Different Online Social Networks

Online social networks have become popular for communication among people. Previous works on social networks were focused mainly on a single network due to the lack of links from one network to the others. There are several ways to connect different social networks. One possible solution is to find out the same individual who owns accounts for different social networks simultaneously. In this paper, we focus on the problem of linking the same users across different social networks. We start from a basic solution based on profile comparison, which is then improved by considering additional social relationships, i.e., friends. Finally, we propose a two-phase clustering method to generate a social summary for each individual. The first phase is to select strongly-connected groups as seeds. The second phase is to assign non-seeds to the clusters based on the social structure and profiles. The final distributions over various clusters are regarded as the social summary. The experiments have been conducted on two real social networking datasets. The experimental results show the feasibility of the proposed (social summary) method, compared to the methods of profile comparison and person-name comparison.