Combining Machine Learning with Shortest Path Methods: Discovering, Visualizing, and Analyzing Hollywood's Power Clusters to Go From Six Degrees of Kevin Bacon to Knowing Colin Firth

摘要

This paper describes a method to model, discover, and visualize communities in social networks. It makes use of a novel method based on the "Six Degrees of Kevin Bacon" principle: find the shortest path between entities in a social graph and then discover communities based on clustering with those shortest-path distances. We have applied this idea to find Hollywood's power clusters based on IMDB (Internet Movie Database), which links actors to movies. Using this method, we found roughly three clusters of Hollywood elite actors, the largest of which contained many of Hollywood's best-known actors. For living actors, we found Colin Firth (who played Pride and Prejudice's Mr. Darcy), Javier Bardem (who played a psychopathic killer in No Country for Old Men), and Joaquin Phoenix (who played Johnny Cash and a Roman Emperor in Gladiator) to be some of the most well-connected actors in Hollywood. This suggests that analyzing a social network using our method can lead to some surprising results.