Privacy Scoring of Social Network User Profiles Through Risk Analysis

The social benefit derived from online social networks (OSNs) can lure users to reveal unprecedented volumes of personal data to a social graph that is much less trustworthy than the offline social circle. Although OSNs provide users privacy configuration settings to protect their data, these settings are not sufficient to prevent all situations of sensitive information disclosure. Indeed, users can become the victims of harms such as identity theft, stalking or discrimination. In this work, we design a privacy scoring mechanism inspired by privacy risk analysis (PRA) to guide users to understand the various privacy problems they may face. Concepts, derived from existing works in PRA, such as privacy harms, risk sources and harm trees are adapted in our mechanism to compute privacy scores. However, unlike existing PRA methodologies, our mechanism is user-centric. More precisely, it analyzes only OSN user profiles taking into account the choices made by the user and his vicinity regarding the visibility of their profile attributes to potential risk sources within their social graphs. To our best knowledge, our work is the first effort in adopting PRA approach for user-centric analysis of OSN privacy risks.