Learning Max-Margin GeoSocial Multimedia Network Representations for Point-of-Interest Suggestion

摘要

With the rapid development of mobile devices, point-of-interestrn(POI) suggestion has become a popular online web service, whichrnprovides attractive and interesting locations to users. In order to providerninteresting POIs, many existing POI recommendation worksrnlearn the latent representations of users and POIs from users’ pastrnvisiting POIs, which suffers from the sparsity problem of POI data.rnIn this paper, we consider the problem of POI suggestion from thernviewpoint of learning geosocial multimedia network representations.rnWe propose a novel max-margin metric geosocial multimediarnnetwork representation learning framework by exploiting users’rncheck-in behavior and their social relations. We then develop arnrandom-walk based learning method with max-margin metric networkrnembedding. We evaluate the performance of our method on arnlarge-scale geosocial multimedia network dataset and show that ourrnmethod achieves the best performance than other state-of-the-artrnsolutions.