An example showing clustering of users. These are groups of users who share similar tastes. Based on the graph you can determine these clusters of users in order to increase the accuracy of recommendations, by weighting the recommendations from someone in the same cluster.

TrackTaste