Personalized PageRank vectors for tag recommendations

摘要: This paper looks inside FolkRank, one of the well-known folksonomy-based algorithms, to present its fundamental properties and promising possibilities for improving performance in tag recommendations. Moreover, we introduce a new way compute differential approach FolkRank by representing it as linear combination personalized PageRank vectors. By combination, FolkRank's probabilistic interpretation that grasps how works on folksonomy graph terms random surfer model. We also propose FolkRank-like methods recommendations efficiently tags' rankings thus reduce expensive computational cost FolkRank. show approaches are feasible recommend tags real-time scenarios well. The experimental evaluations proposed provide fast with reasonable quality, compared Additionally, discuss diversity top n recommended variants.