Link Prediction in the Greek Web

In this notebook we are dealing with the problem of link prediction in graphs.Specifically we have a subset of the Greek web and we are trying to predict missing edges. The approach we are going to follow is to deal with the problem as a binary classification task, where each edge is a candidate new edge for the graph. In the specific dataset there are 2041 nodes and 2683 edges and we are trying to predict the 453 edges that have been manually removed from the graph. We also have a dataset of different texts available from the hosts.