Inferring privacy information via social relations

Currently, millions of individuals are sharing personal information and building social relations with others, through online social network sites. Recent research has shown that those personal information could compromise owners´ privacy. In this work, we are interested in the privacy of online social network users with missing personal information. We study the problem of inferring those users´ personal information via their social relations. We present an iterative algorithm, by combining a Bayesian label classification method and discriminative social relation choosing, for inferring personal information. Our experimental results reveal that personal information of most users in an online social network could be inferred through mere social relations with high accuracy.