Learning-based link prediction analysis for Facebook100 network

In social network science, Facebook is one of the most interesting and widely used social networks and media platforms. In the previous decade Facebook data contributed to significant evolution of social network research. Paired with this topic we have experienced growing popularity in the link prediction techniques, which are important tools in link mining and analysis. This paper gives a comprehensive overview of link prediction analysis on the Facebook100 network, which was derived in 2005. We study performance and evaluate multiple machine learning algorithms on this network. We use networks embeddings and topology-based techniques such as node2vec and vectors of similarity metrics. Using these techniques similarity features for our classification models are derived. Further we discuss our approach and present results. Lastly, we compare and review our models, where overall performance and classification rates are presented.