PriGuardTool: A Web-Based Tool to Detect Privacy Violations Semantically

Online social networks contain plethora of information about its users. While users enjoy sharing information online, not all information is meant to be seen by the entire network. Managing the privacy of users has become an important aspect of such online networks. An important part of this is detecting privacy violations and notifying the users so that they can take appropriate actions. While various approaches for detecting privacy violations exist, most of the approaches do not have a running tool that can exhibit the principles of its underlying approach. This paper presents PriGuardTool, a Web-based tool that can detect privacy violations in online social networks. Each user is represented by a software agent in the system that first collects user’s privacy concerns, explicitly specified as what types of content are meant to be seen by which audience. The system represents these privacy constraints as commitments between the user and the online social network. The user constraints are converted into commitments automatically by the agent. The system then monitors which commitments are violated based on the content shown to users, such that a violated commitment represents a privacy violation in the system. While checking for violations, the effects of posts on the system as well as the semantic relations and rules are considered. We evaluate PriGuardTool by using various real-life scenarios and real data that have been collected over Facebook. Our initial results show that realistic privacy violations can be detected using PriGuardTool.