As part of the Netflix Prize competition, chef-ele has created a computer algorithm that can show different clusters of movies within the Netflix database. Chef-ele has used the open-source graph visualization package GUESS for most of his visualizations.
As he explains: "To generate those graphs I basically calculated the full movie-movie correlation matrix, using a straight Pearson correlation with Fisher confidence interval correction & a few other tweaks. Then my program writes a textfile of node & edge definitions, which GUESS reads in. Each movie is a node in the graph, with a label (movie title). The lines/edges connect each movie with the one most similar neighbor movie (in the Fisher/Pearson sense). (...) The graphs I did make show some very reasonable clusters. Some movies -- especially those in series & sequels -- have stronger / larger clusters (such as Friday the 13th sequels the 'Friends' TV series, everything Star-Trek related, etc). But clusters around 'themes' (kids movies, shoot-em-ups, etc.) were also found by the algorithm & seem pretty clear."