Date of Award

Fall 10-1-2021

Document Type


Degree Name

Doctor of Philosophy (PhD)


Computer Science

First Advisor

Feigenbaum, Joan


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. This dissertation addresses both goals. First, we consider the task of {\it interorganizational data sharing}, in which data owners, data clients, and data subjects have different and sometimes competing privacy concerns. A key challenge in this type of scenario is that each organization uses its own set of proprietary, intraorganizational attributes to describe the shared data; such attributes cannot be shared with other organizations. Moreover, data-access policies are determined by multiple parties and may be specified using attributes that are not directly comparable with the ones used by the owner to specify the data. We propose a system architecture and a suite of protocols that facilitate dynamic and efficient interorganizational data sharing, while allowing each party to use its own set of proprietary attributes to describe the shared data and preserving confidentiality of both data records and attributes. We introduce the novel technique of \textit{attribute-based encryption with oblivious attribute translation (OTABE)}, which plays a crucial role in our solution and may prove useful in other applications. This extension of attribute-based encryption uses semi-trusted proxies to enable dynamic and oblivious translation between proprietary attributes that belong to different organizations. We prove that our OTABE-based framework is secure in the standard model and provide two real-world use cases. Next, we turn our attention to utility that can be derived from the vast and growing amount of data about individuals that is available on social media. As social networks (SNs) continue to grow in popularity, it is essential to understand what can be learned about personal attributes of SN users by mining SN data. The first SN-mining problem we consider is how best to predict the voting behavior of SN users. Prior work only considered users who generate politically oriented content or voluntarily disclose their political preferences online. We avoid this bias by using a novel type of Bayesian-network (BN) model that combines demographic, behavioral, and social features. We test our method in a predictive analysis of the 2016 U.S. Presidential election. Our work is the first to take a semi-supervised approach in this setting. Using the Expectation-Maximization (EM) algorithm, we combine labeled survey data with unlabeled Facebook data, thus obtaining larger datasets and addressing self-selection bias. The second SN-mining challenge we address is the extent to which Dynamic Bayesian Networks (DBNs) can infer dynamic behavioral intentions such as the intention to get a vaccine or to apply for a loan. Knowledge of such intentions has great potential to improve the design of recommendation systems, ad-targeting mechanisms, public-health campaigns, and other social and commercial endeavors. We focus on the question of how to infer an SN user's \textit{offline} decisions and intentions using only the {\it public} portions of her \textit{online} SN accounts. Our contribution is twofold. First, we use BNs and several behavioral-psychology techniques to 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). Second, we explore the extent to which temporal models may assist in the inference task by representing SN users as sets of DBNs that are built using our modeling techniques. The use of DBNs, together with data gathered in multiple waves, has the potential to improve both inference accuracy and prediction accuracy in future time slots. It may also shed light on the extent to which different factors influence the decision-making process.