NodeSim: node similarity based network embedding for diverse link prediction

Abstract In real-world complex networks, understanding the dynamics of their evolution has been great interest to scientific community. Predicting non-existent but probable links is an essential task social network analysis as addition or removal over time leads evolution. a network, can be categorized intra-community if both end nodes link belong same community, otherwise inter-community links. The existing link-prediction methods have mainly focused on achieving high accuracy for prediction. this work, we propose embedding method, called NodeSim, which captures similarities between and community structure while learning low-dimensional representation network. learned using proposed NodeSim random walk, efficiently explores diverse neighborhood keeping more similar closer in context node. We verify efficacy method state-of-the-art machine model prediction that considers nodes’ information predict two given nodes. Extensive experimental results several networks demonstrate effectiveness inter