A Survey of Different Methods in Finding Latent Relationships among Complex Networks

Predicting relationships or links in a network is one of the major problems in computer science and has found applications in a variety of complex networks. In the prediction of link in a highly connected or sparse network at a time w1, we have to find the probability of occurrence of the missing links or links that might manifest in the future at a new point of time w1+1. It can also be used to remove spurious links or noisy links in the network that might have cropped up as the complexity of the network is very high. In this article, we have tried to summaries four different approaches to solving link prediction problem which is based on similarity measures from nodes in a graph. These approaches have been developed recently and have taken their inspiration from various fields of study. Finally, we have presented the results, preprocessing approach and evaluation metrics used in order to compare these new techniques, also we have mentioned future challenges and applications of the link prediction algorithm.