A novel collaborative filtering approach based on social network experts
Yasser El Madani El Alami, El Habib Nfaoui, Omar El Beqqali · 2016
Collaborative filtering is considered as the most popular approach for generating recommendations. It provides various techniques which rely on similar users (or items) to compute recommendations. In order to solve many problems such as sparsity, scalability and shilling attacks, there has been a great interest to expert-based collaborative methods. However, selecting reliable experts is still under study. In this work, we investigate the effect of incorporating the social weight of experts in the selection phase. We present a variant definition of an expert which takes into account its centrality weight in the social graph. Second we propose our collaborative filtering algorithm based on experts using an adjusted similarity measure. Finally, experiments show that our approach provides better performance than the other tested methods.