Tell Me What You Want and I Will Tell Others Where You Have Been
Anthony Quattrone, Elham Naghizade, Lars Kulik, Egemen Tanin · 2014
Trajectory data does not only show the location of users over a period of time, but also reveals a high level of detail regarding their lifestyle, preferences and habits. Hence, it is highly susceptible to privacy concerns. Trajectory privacy has become a key research topic when sharing/exchanging trajectory datasets. Most existing studies focus on protecting trajectory data through obfuscating, anonymising or perturbing the data with the aim to maximize user privacy. Although such approaches appear plausible, our work suggests that precise trajectory information can be inferred even from other sources of data. We consider the case in which a location service provider only shares POI query results of users with third parties instead of exchanging users' raw trajectory data to preserve privacy. We develop an inference algorithm and show that it can effectively approximate original trajectories using solely the POI query results.