A Novel Collaborative Filtering Algorithm Based on Bipartite Network Projection

In this paper, we present a collaborative filtering algorithm based on the bipartite network projection. First, we project the bipartite network to a weighted directed one-mode network in order to retain original information and extract hidden information. The weight of an edge denotes how important the ending vertex is to the starting one. Then, we implement the algorithm using projection networks. Based on a benchmark database, numerical results show that our novel algorithm can further improve the algorithmic accuracy and generate more diverse and less popular recommendation.