Protecting Hidden Location Visit Privacy with High Trajectory Data Utility

Abstract

With the spread of social network applications with check-in service, people's trajectories are continuously recorded. The trajectory data are often published for personalized recommendations and activity mining. However, publishing trajectory data makes users' hidden location visits vulnerable to inference attacks. These hidden location visits could be considered as leaked, if an adversary can infer them with a high confidence. In this paper, we study the problem of protecting hidden location visits in the publication of trajectory data, assuming an adversary can do inference attacks on trajectory data. We propose a hidden location visit privacy protection algorithm employing location replacement and location suppression, to protect hidden location visits against inference attacks. We also design a number of techniques to make the anonymized trajectory data match users' behavior patterns. The experimental results show that our algorithms can efficiently prevent inference attacks on real datasets while preserving high trajectory data utility.