Information Filtering Based on Personalized Topology Information

In the recommendation of the bipartite networks, researchers have mainly dedicated to improve the accuracy of the recommendation, but neglected the fact that the entire history information which can be redundant or even misleading to the performance of recommendation. In this paper, we set unique weight to every link according to their temporal information and topology information. Then, we remove the links according to their weight. Experimental results on a number of real networks show that the algorithm improves the recommendation accuracy, meanwhile, largely shortens the computational time complexity and reduces the data storage space.