Top Location Anonymization for Geosocial Network Datasets

Geosocial networks such as Foursquare have access to users' location information, friendships, and other potentially privacy sensitive information. In this paper, we show that an attacker with access to a naively-anonymized geosocial network dataset can breach users' privacy by considering location patterns of the target users. We study the problem of anonymizing such a dataset in order to avoid re-identification of a user based on her or her friends' location information. We introduce k-anonymity-based properties for geosocial network datasets, propose appropriate data models and algorithms, and evaluate our approach on both synthetic and real-world datasets.