Self-adaptive Privacy Concern Detection for User-Generated Content

To protect user privacy in data analysis, a state-of-the-art strategy is differential which scientific noise injected into the real analysis output. The masks individual’s sensitive information contained dataset. However, determining amount of key challenge, since too much will destroy utility while little increase risk. Though previous research works have designed some mechanisms to different scenarios, most existing studies assume uniform concerns for all individuals. Consequently, putting an equal individuals leads insufficient protection users, over-protecting others. address this issue, we propose self-adaptive approach concern detection based on personality. Our experimental demonstrate effectiveness suitable personalized cold-start users (i.e., without their privacy-concern training data).