Supervised Learning for Link Prediction in Social Networks

Link prediction is an estimation problem that has drawn a great deal of attention in recent years. In this work, a supervised learning approach is adopted to perform link prediction on data retrieved from Facebook. The specific goal, then, is to estimate the probability of two users to become friends in order to recommend them to one another whenever this probability turns out to be sufficiently high. On social platforms like Facebook, friendship recommendation is clearly a crucial ingredient since, when properly implemented, it plays a key role in determining the network growth. The contribution of this work consists in performing friendship recommendation on Facebook using a supervised learning approach that takes explicitly into account vertices’ attributes; that is, all the personal information that users make available on their profiles.