Privacy challenges in the online social networking era.

ABSTRACTOnline Social Networking (OSN) services such as Facebook and Twitter are immensely popular. Their users entrust them with sensitive data such as friends lists, pictures and messages. This data can be directly shared with other users or can be handed to the third-party applications for further processing. Our work focuses on on the many privacy and trust issues that OSNs present. More specifically, we are interested in studying users' privacy in OSNs under various trust and attack models. To deal with untrusted service providers, we propose two decentralized frameworks for building privacy-preserving OSN platforms in which access to privacy-sensitive information is limited to (1) user-controlled code hosted by "infrastructure as a service" (IaaS) providers such as EC2, and (2) desktop PCs controlled by a user's close friends. We also designed a secure "platform-as-a-service" (PaaS) capable of ensuring that hosted third-party applications adhere to users' privacy policies. Finally, we consider designing a secure protocol that uses properties of a social graph for matching mutually distrustful users without undermining their privacy.