I can infer where you live: a location prediction approach using Facebook profile public information

7th May 2015

Ángel Cuevas Rumín , assistant professor, Universidad Carlos III de Madrid, Madrid, Spain.

Abstract: I can infer where you live: a location prediction approach using Facebook profile public information On-line social networks leak a substantial amount of private information from the users registered in those systems. Although a user can proactively protect some private item (e.g., location, employment, interests, etc) by making them private in their OSN profile, this does not guarantee that this information cannot be uncovered using: (i) other profile attributes, (ii) public information available in the OSN profiles of friends. In this talk, we first analyze the amount of information publicly disclosed in Facebook. Then, we propose a model that very accurately obtains the "Current City" (i.e., user location) for Facebook users that have hidden that information. Finally, we rely on the previous model to propose a "Current City Exposure Estimator" that informs Facebook users of the probability that a third party can retrieve their "Current City" information even in the case it is not public in their profile. This estimator is used to inform Facebook users on what countermeasures they could take to reduce their "Current City" exposure.