Privacy-Preserving Social Tie Discovery Based on Cloaked Human Trajectories

Abstract

© 2016 IEEE. The discovery of peoples' social connections is becoming a flourishing research topic, considering the rich social information inferable from human trajectories. Existing social tie detection methods often require mobile users to upload their accurate locations, causing serious privacy concerns. On the other hand, cloaking methods allow users to upload their obscured locations instead and can efficiently protect their location privacy. However, no existing social tie detection method can generate social relationships among users when only obscured trajectories are provided. To tackle the aforementioned problem, this paper proposes a novel semantic-tree-based algorithm. Specifically, we model the obscured regions from the cloaking algorithm as a semantic region tree and assign weight values for regions based on their popularity, further indicating the similarity between users based on their temporal and spatial relations. We evaluate our proposed approach using a real trajectory data set and show that our algorithm can identify social ties successfully with 20% higher accuracy than the existing approaches.