An improved link prediction algorithm based on common neighbors index with community membership information

human life is more and more dependent on the safety, reliability and effective operation of a variety of complex networks; however, most of networks are sparse, which means the network data is incomplete. To solve the problem, various link prediction methods have been proposed to find missing links in given networks. Among these methods, similarity-based methods are effective, however, still imperfect. In order to improve the predicting results, we combined local and global information of the network and then proposed a method based on one of similarity-based methods with community information. Experiments results show that the inclusion of community information improves the accuracy of results of predicting missing links.