Detecting and predicting privacy violations in online social networks

摘要: Online social networks have become an essential part of and work life. They enable users to share, discuss, create content together with various others. Obviously, not all is meant be seen by all. It extremely important ensure that only shown those are approved the content's owner so owner's privacy preserved. Generally, online promising preserve through agreements, but still everyday new leakages taking place. Ideally, should able manage maintain their agreements well-founded methods. However, dynamic nature making it difficult keep private information contained. We developed $\mathcal{PROTOSS}$ , a run time tool for detecting predicting $\mathcal{PR}\mathrm{ivacy}\ \mathrm{vi}\mathcal{O}\mathrm{la}\mathcal{T}\mathrm{ions}\ \mathrm{in}\ \mathcal{O}\mathrm{nline}\ \mathcal{S}\mathrm{ocial}\ \mathrm{network}\mathcal{S}$ . captures relations among users, network operator, as well domain-based semantic rules. uses model checking detect if will result in violation agreements. can further use infer possible violations been specified user explicitly. In addition detection, predict future feeding hypothetical world state. Through running example, we show subtle leakages, similar ones reported real life examples. We study performance our system on scenario existing Facebook dataset.