Combining Factorization Model and Additive Forest for Collaborative Followee Recommendation

Social networks have become more and more popular in recent years. This popularity creates a need for personalization services to recommend tweets, posts (information) and celebrities organizations (information sources) to users according to their potential interest. Tencent Weibo (microblog) data in KDD Cup 2012 brings one such challenge to the researchers in the knowledge discovery and data mining community. Compared to traditional scenarios in recommender systems, the KDD Cup 2012 Track 1 recommendation task raises several challenges: (1) Existence of multiple, heterogeneous data sources; (2) Fast growth of the social network with a large number of new users, which causes a severe user cold-start problem; (3) Rapid evolution of items’ popularity and users’ interest. To solve these problems, we combine feature-based factorization models with additive forest models. Specifically, we first build factorization models that incorporate users’ social network, action, tag/keyword, profile and items’ taxonomy information. Then we develop additive forest models to capture users’ activity and sequential patterns. Because of the additive nature of such models, they allow easy combination of the results from previous factorization models. Our modeling approach is able to utilize various side information provided by the challenge dataset, and thus alleviates the cold-start problem. The new temporal dynamics model we have proposed using an additive forest can automatically adjust the splitting time points to model popularity evolution more accurately. Our final solution obtained an MAP@3 of 0.4265 on the private leader board, giving us the first place in Track 1 of KDD Cup 2012.