Title: Data exploitation and privacy protection in the era of data sharing.
Advisor: Joan Feigenbaum
Other committee members:
Bryan Ford (EPFL)
As the amount, complexity, and value of data available in both private and public sectors has risen sharply, the competing goals of data privacy and data utility have challenged both organizations and individuals. My work addresses both goals.
First, I will present a novel Bayesian-network (BN) approach to the mining of publicly observable social-network data. The effectiveness of the method is demonstrated using multiple models that aim at inferring the voting behavior of social-network users in the 2016 US presidential election. When enhanced with dynamic BNs that model decision making as a complex process that both influences and is influenced by static factors (such as personality traits and demographic categories) and dynamic factors (such as triggering events, interests, and emotions), this method can be used to infer social-network users’ intentions regarding actions that they have not yet taken in areas such as health and finance.
Next, I will present PRShare, a system that I have designed and implemented and that enables efficient, privacy-preserving interorganizational data sharing. PRShare makes essential use of the novel cryptographic technique of attribute-based encryption with oblivious attribute translation, an extension of attribute-based encryption that supports expressive decryption policies for both data and metadata attributes.