Friendship and affiliation co-recommendation via Collective Latent Factor BlockModel

With the increasing of online social networks, people always form the friendship networks among their social neighborhood, and also associate themselves with circles or communities due to their common interest. Thus there are two related networks: the friendship network among users as well as the affiliation network between users and circles. In this paper, we address the problem of collaborative recommendation for friendships and affiliations in online social networks. For that we propose the Collective Latent Factor BlockModel (CLFBM) to collectively discover globally predictive intrinsic properties of users and capture the interpretable latent block structure corresponding to the circle information. The proposed model is exploited in a transfer learning framework that extracts knowledge from the two related networks and generates recommendations for friendships and affiliations. The extensive experiments on the real world datasets suggest that our proposed CLFBM model outperforms the other state of the art approaches in the recommendation tasks.