On the Privacy Implications of Location Semantics

International audienceMobile users increasingly make use of location-based on-line services enabled by localization systems (e.g., GPS). Not only do they share their locations to obtain contextual services in return (e.g., 'nearest restaurant'), but they also share information about the venues (e.g., the type, such as a restaurant or a cinema) they visit with their friends. This introduces an additional dimension to the threat to location privacy: location semantics that, combined with location information, can be used to improve location inference by learning and exploiting patterns at the semantic level (e.g., people go to cinemas after going to restaurants). Conversely, the type of venue a user visits can be inferred, hence this knowledge can be used to aggravate the threat to her (semantic) location privacy. In this paper, we formalize this problem and analyze the effect of venue-type information on location privacy. We introduce two multidimensional inference models that consider location semantics and semantic privacy-protection mechanisms. We evaluate our models by using a real data-set of semantic check-ins from Foursquare (obtained through Twitter). Our experimental results show that users' semantic location privacy is at serious risk and that semantic information significantly improves inference of user locations, hence degrading location privacy