Scalable Approaches for Recommendation in Social Networks

Yifan Li (1)

1 BD - Bases de Données

Nowadays, the online social network has become a significant part of our life, and deeply influenced our activities in many aspects than one might imagine before. Hereby, with the benefit from its increasing growing, there are more prediction opportunities emerging for recommendation system deployment. Rather than as only a supplementary of the traditional use of collaborative filtering (CF) method, in some cases[7], the social affinity among users can provide more precision in recommendation result(scores) calculation. In this doctoral thesis, we propose to design a scalable recommendation approach over large social graph, taking into consideration not only graph topological properties but also those user semantic content, e.g. user interest, (hash-)tags, item ratings, to gain a good and fast recommendation evaluation.

Mots clés: Social Network; Recommendation System; Graph; Pregel; Personalized PageRank

Domaines: Base de données [cs.DB]

Yifan Li. Scalable Approaches for Recommendation in Social Networks. BDA2015, Sep 2015, Toulon, France. ⟨hal-01398199⟩