Learning recommendations in social media systems by weighting multiple relations

摘要: We address the problem of item recommendation in social media sharing systems. adopt a multi-relational framework capable to integrate different entity types available system and relations between entities. then model tasks as weighted random walks relational graph. The main contribution paper is novel method for learning optimal each relation given task, by minimizing loss function on training dataset. report results weight two common Flickr dataset, tag images contact users.