A Crowdsourced Privacy Preserving Social Platform Driven by Users’ Consensus

Abstract

Privacy enforcement systems in social platforms often neglect to capture the sophisticated, ever-shifting privacy expectations of their users. The theory of contextual integrity (CI) seeks to eliminate this discrepancy by transforming privacy expectations into norms and allowing only appropriate information flows based on these norms. The derivation, encoding, and enforcement of the large number of norms required pose a main challenge to implementing a CI based social platform. To address these difficulties, and to design a more scalable system, we propose a crowdsourced norm generation process in which the users, rather than the privacy experts, collectively decide which norms to approve. The proposed system then encodes norms and information flows using the logic programming language Datalog, which also serves as the enforcement mechanism for the platform website. To evaluate the feasibility of crowdsourced norm derivation, we conducted extensive surveys for over 2000 norms in the educational context involving over 700 participants. Based on the responses, we arrived at a widely approved subset that not only demonstrates the framework’s ability to derive norms but also provides a starting point to bootstrap the system. We then used the Z3 theorem prover to ensure non-conflicting encodings and remove possible logical inconsistencies from the approved subset. As a whole, our work proposes a scalable and adaptable approach to designing and implementing a CI based social platform.