Identifying vulnerable friends on a social networking site

As social network is becoming widely used, the problem of protection of user´s privacy has caused heated debate. It is necessary to let users know who would propagate or leak their personal information, and we define those people as vulnerable friends. Most existing approaches are either personal information based or users´ influence based. In this paper, we aim to mine the vulnerable information behind, and thus better identify vulnerable friends. In this paper, we present a Fuzzy Analytical Hierarchy Process (FAHP) based method by fusing users´ influence consideration. Apparently, multiple factors (e.g., gender, birthday, hometown, mobile phone, high school, college etc.) may affect propagation of privacy information. Therefore, we first present the calculation of users´ influence. Second, we adopt FAHP to analysis users´ information and consider their probability of propagating other people´s information. At last, we use Coefficient of Variation combine those factors. A thorough experimental comparison, based on the data which regards to Facebook, illustrates advantage of our scheme over other algorithms.