Generalized Stochastic Block Model
[Jamali et al. 2011]
• Social influence and selection lead to
formation of communities/groups.
• Users belong to different (latent) groups,
e.g. teacher interacting with students or his/her son
or camera being rated by professional vs. amateur.
• Items belong to different (latent) groups,
e.g. high-quality and low-quality items.
• Clustering based method for recommendation.
Significance of friend recommendation
– All major social networks have it.
– E.g., in LinkedIn 50% of connections from
recommendations. [Posse 2012]
C. Posse: Key Lessons Learned Building Recommender Systems for Large-Scale Social Networks,
KDD 2012.
A. Sharma, D. Cosley: Do social explanations work? Studying and modeling the effects of social
explanations in recommender systems, WWW 2013.
Supervised random walks [Backstrom & Leskovec 2011
Model based Methods
• MF based models [Rennie & Srebo 2005]
– Social network as a binary matrix.
– Similar to MF methods for rating prediction.
– Factorize the network matrix into product of
lower rank matrices (representing user factors).
– Advanced version in [Yang et al. 2011]:
factorize user-user matrix and user-item matrix
simultaneously.
Bo Hu, Martin Ester: Spatial Topic Modeling in Online Social Media for Location Recommendation. RecSys 2013
Samaneh Moghaddam, Martin Ester: The FLDA model for aspect-based opinion mining: addressing the cold start problem. WWW 2013: 909-918
Bo Hu, Zhao Song, Martin Ester: User Features and Social Networks for Topic Modeling in Online Social Media. ASONAM 2012: 202-209
Bo Hu, Mohsen Jamali, Martin Ester: Learning the Strength of the Factors Influencing User Behavior in Online Social Networks. ASONAM 2012: 368-375