Toward preserving privacy and functionality in geosocial networks

Abstract

Storing user friend lists, preferences and messages, online social networks have become a significant source of sensitive personal information. A recent addition to this space, geosocial networks (GSNs) such as Yelp [1] or Foursquare [2], collect even user locations, through check-ins performed by users at visited venues. Overtly, personal information allows GSN providers to offer a variety of applications, including personalized recommendations and targeted advertising, and venue owners to promote their businesses through spatio-temporal incentives (e.g., rewarding frequent customers through accumulated badges). Providing personal information exposes however users to significant risks, as social networks have been shown to leak [3] and even sell [4] user data to third parties. There exists therefore a conflict. Without privacy people may be reluctant to use geosocial networks; without user information the provider and venues cannot support applications and have no incentive to participate. In this work we take first steps toward breaking this deadlock, by introducing the concept of location centric profiles (LCPs), aggregate statistics built from the profiles of users that have visited a certain location. As we know, location privacy has been extensively studied before [5]. This work significantly extends the state of the art by (i) providing constructs that preserve the privacy of users when reporting private profile information (e.g., age, gender, location), and (ii) ensuring that the solutions enable providers to collect information needed to develop existing services. We introduce ProfilR , a framework that allows the construction of LCPs based on the profiles of present users, while ensuring the privacy and correctness of participants. To relieve the GSN provider from costly involvement in venue specific activities, ProfilR stores and builds LCPs at venues.