An Improved Collaborative Filtering Algorithm Based on Node2vec
Liang Luo, Ruichun Tang · Proceedings of the 2018 2nd International Conference on Computer Science and Artificial Intelligence · 2018
Collaborative filtering (CF) algorithms are widely used in online recommendation systems, which can predict user's preference based on their historical behavior. However, traditional methods confront the challenges of sparsity and efficiency due to the growth of users and items. This paper proposed a novel method which incorporates social latent influence into CF to address above issues. In brief, we utilize node2vec technology to learn user complex latent relationship among users from the social network and then apply this latent relationship and historical rating to find the neighbors of target users. The experiments on Last.FM and Sobazaar datasets show the superiority of the method in regard to precision, recall, and F-measure.