Performing link prediction on graphs using graph embeddings technique generated by the NBNE (https://github.com/tpimentelms/nbne).

The graph used in the example was retrieved from the page "https://snap.stanford.edu/data/egonets-Facebook.html", by downloading the file "facebook_combined.txt.gz".

This is a sample of graph learning on how to perform link prediction using an unsupervised learning technique to obtain dense vectorial representations for graph vertices on (extremely) imbalanced graphs (there are plenty more non-endges than edges in the target graph).

Two models are analysed here, a XGBoost and a Transformer Neural Network.