Netflix Recommendations
In the world of Netflix, roughly 75 percent of what people are watching is influenced by its recommendations.The algorithm for generating these recommendations incorporates even the most obscure factors such as the type of device used for viewing, and even the time of the day. The structure of how movies and TV shows are recommended can be represented by a network. If you think of a TV show or movie as the nodes, common tags between the shows form the edges. The more tags two shows or movies share, the stronger the tie between them.
The network representing the relationship between TV shows and films have bridges and triadic closures. Say Netflix recommends a movie of a genre that you do not watch, like a documentary. This can happen if you liked a comedy movie (a genre you frequently watch) that was directed by someone that happened to direct that one documentary in his career that is unrelated to anything you’ve ever watched. This recommendation was most likely based on an edge that serves as a bridge between genres, with that edge being the common director. Triadic closures are incorporated in this network in the sense that if Film A is at the top of Film B’s recommended list, and Film C is at the top of Film A’s recommended list, it is likely that Film C will be somewhere on Film B’s recommended list, since if A and B share a lot of tags, and B and C share a lot of tags, it is likely that A and C have a decent amount of common tags as well. With Netflix’s strong influence of recommendations on their viewers, the network relating its shows and movies to each other basically controls what you’ll be binge watching next.
http://www.wired.com/2013/08/qq_netflix-algorithm/