Fine-grained preference-aware location search leveraging crowdsourced digital footprints from LBSNs

Abstract

The crowdsourced digital footprints from Location Based Social Networks (LBSNs) contain not only rich information about locations, but also individual's feeling about locations and associated entities. This new data source provides us with an unprecedented opportunity to massively and cheaply collect location related information, and to subtly characterize individual's fine-grained preference about those places and associated entities. In this paper, we propose SEALs - a fine-grained preference-aware location search framework leveraging the crowdsourced traces in LBSNs. We first collect user check-ins and tips from Foursquare and use them as direct user feedback on locations. Second, we extract users' sentiment about locations and associated entities from tips to characterize their fine-grained location preference. Third, we incorporate such fine-grained user preference into personalized location ranking using tensor factorization techniques. Experimental results show that SEALs can achieve better location ranking comparing to the state-of-the-art solutions.