Spotting misbehaviors in location-based social networks using tensors

The proliferation of mobile devices that are capable of estimating their position, has lead to the emergence of a new class of social networks, namely location-based social networks (LBSNs for short). The main interaction between users in an LBSN is location sharing. While the latter can be realized through continuous tracking of a user's whereabouts from the service provider, the majority of LBSNs allow users to voluntarily share their location, through check-ins. LBSNs provide incentives to users to perform check-ins. However, these incentives can also lead to people faking their location, thus, generating false information. In this work, we propose the use of tensor decomposition for spotting anomalies in the check-in behavior of users. To the best of our knowledge, this is the first attempt to model this problem using tensor analysis.