Biega, Joanna Asia, et al. "R-susceptibility: An ir-centric approach to assessing privacy risks for users in online communities." Proceedings of the 39th International ACM SIGIR conference on Research and Development in Information Retrieval. 2016.

This paper introduces a novel ranking-based methodology for evaluating privacy risks associated with user-generated textual content in online communities, specifically focusing on sensitive subjects like mental health issues. The approach models an adversary using latent topic models to understand vocabulary and correlations, aiming to identify users most at risk of being targeted. The proposed R-Susceptibility model aims to quantify privacy risks, allowing users to be informed and alerted about their vulnerability to adversarial actions, such as discriminatory pricing or targeted advertising of sensitive nature. Experimental results using real-world data demonstrate the practicality of this approach in assessing and mitigating privacy risks in online communities.