Towards usable differentially private analyses — Exploring suitable metaphors for lay users


In this presentation, we present our work, funded by PAPAYA and TRUEdig projects, on the suitability of the metaphors for aiding informed decisions of data subjects on sharing their data with DP systems and discuss open research challenges. Despite the recent enhancements in the deployment of differential privacy (DP), little research has been conducted addressing the human aspects of DP-enabled systems. Metaphors could be a suitable means for conveying key protection functionalities of DP to lay users. We extracted and generated metaphors for local and central differentially private data analysis models. We first analytically evaluated the metaphors based on experts’ feedback followed by an empirical evaluation via online interviews with 30 participants. The results of our interviews show that the metaphorical explanations can successfully convey privacy features of DP such as e.g. that perturbation protects privacy and that there is a privacy-accuracy trade-off. However, conveying information at a high level limits the ability to apply and extend the concept to different contexts and leads to wrong expectations which negatively affects their understanding. Challenges for usable differential privacy and ways for approaching them are discussed.

Mar 11, 2022 1:15 PM — 2:15 PM
Room EDIT 8103, Campus Johanneberg
Chalmers University of Technology
Rännvägen 6B, Gothenburg, 412 58

Simone Fischer-Hübner received the Diploma degree in computer science (law), in 1988, and the Ph.D. and Habilitation degrees in computer science from the University of Hamburg, Germany, in 1992 and 1999, respectively. She has been a Full Professor with Karlstad University, Sweden, since 2000, where she is currently the Head of the Privacy and Security Research Group. She is also a Scientific Coordinator with the EU H2020 Marie Skłodowska-Curie ITN Privacy & Us. She has contributed as a Partner with the CyberSec4Europe, PAPAYA, CREDENTIAL, PRISMACLOUD, A4Cloud, SmartSociety, PrimeLife, PRIME, FIDIS, and Bugyo EU projects. Her research interests include cyber security, privacy-enhancing technologies, and usable privacy and security. She is a Swedish IFIP TC 11 Representative and a member of the Advisory Board Swedish Civil Contingency Agency’s Cyber Security Council. She serves as the Vice Chair for the IEEE Sweden Computer/Software Engineering Chapter.

Farzaneh Karegar has a Bachelor’s degree from University of Tehran (UT), a Master’s degree in Computer Engineering from Shahid Beheshti University (SBU) and a PhD from Karlstad University. Currently, she is working on algorithmic transparency and usable transparency of privacy-preserving data analytics with a focus on usable differential privacy.