Interchange-Based Privacy Protection for Publishing Trajectories

Abstract

Information extracted from trajectory data is very useful in many practical application scenarios. Before trajectories for data mining are published, they need to be processed to protect the privacy of the trajectories' bodies. In this paper, a method for such privacy protection is proposed. Our method guarantees that the generated trajectory points satisfy the k-anonymity by interchanging the positions of the trajectory points on the k-core subnet of the relation network. The method treats the trajectory points as the privacy protection object. It overcomes the curse of dimensionality resulting from the K-anonymity of trajectories, and reduces the distortion of the generated trajectories significantly. Moreover, our proposed strategy can preserve the original positions of the trajectory points. Experiments on both real-life and synthetic data sets are carried out with different methods for comparison. The results show that our method has greater efficiency and lower distortion of the processed trajectories.