




Abstract:
Considerable advances in addressing privacy threats have been made in recent years in terms of computer and network security, user control mechanisms, privacy policies and ethical considerations; however, changes in the technological environment are creating numerous new and unaddressed risks to user privacy. In particular, ubiquitous social applications raise new privacy risks which we call social inferences. Social inferences are inferences that result from using social applications and are about user-specific information associated with these applications, such as identity, location, activities and profile information. While numerous social computing applications aim to address privacy concerns through access control, direct access control models fail to prevent unwanted inferences unless they incorporate inference management. Unfortunately, previous inference prevention methods cannot adequately address social inferences because: 1) such inferences don’t result from logical deductions and associations; 2) users’ background knowledge is a premise for most social inferences; 3) the sensitivity of user information may have a dynamic nature based on context, such as time and location.
In this dissertation proposal information entropy is used to predict social inferences by estimating and modeling users’ background knowledge. User experiments are used to confirm the theory, investigate users’ background knowledge, and determine the dynamics of handling privacy thresholds. Challenges of applying the theory in practice, such as the computational complexity are also explored and a system architecture is proposed. The applicability of automatic and semi-automatic methods is evaluated through simulations.
Committee Members:
Dr. Sotirios Ziavras, Professor, ECE Department, NJIT (Advisor)
Dr. Quentin Jones, Associate Professor, IS Department, NJIT, (Co-advisor)
Dr. Nirwan Ansari, Professor, ECE Department, NJIT
Dr. Mengchu Zhou, Professor, ECE Department, NJIT
Dr. Roberto Rojas-Cessa, Associate Professor, ECE Department, NJIT
Click here for seminar archive
Note: All MS thesis and PhD dissertation (proposal) defense are counted towards ECE791.



