Collaborative Privacy Management with Auctioning Mechanisms

Online social networks enable users to share content with other users. Many times, a shared content, such as a group picture, may reveal private information about the uploader as well as others who are associated with the content. Ideally, protection of privacy in such cases would need to consider the privacy concerns of all relevant individuals. However, these concerns might conflict and satisfying one user’s privacy needs could cause a privacy violation for others. This calls for computational mechanisms that can decide on the privacy policies of the content collaboratively. Accordingly, we propose an agent-based collaborative privacy management model for online social networks (OSNs). Agents represent OSN users and manage their privacy requirements on their behalf. We extend Clarke-Tax mechanism for auctioning to achieve fair handling of privacy settings and to tax the agents whose privacy settings are chosen. We evaluate our approach over multi-agent simulations and show that it produces privacy policies efficiently and more accurately than existing approaches.