A new similarity-based link prediction algorithm based on combination of network topological features

In recent years, the study of social networks and analysis these in various fields have grown significantly. One most widely used is issue link prediction, which has recently been very popular among researchers. A a network means communication between members network, can include friendships, cooperation, writing joint article or even membership common place such as company club. The main purpose prediction to investigate possibility creating deleting links future state using its current state. this paper, three new similarities, degree neighbor similarity (DNS), path (PNS) Similarity (DPNS) criteria are introduced neighbor-based path-based criteria, both use graph structures. results tested based on area under curve (AUC) precision datasets it shows well superiority work over that only path.