In 2006 Netflix, a DVD rental company, organized the famous Netflix competition. The goal of this competition was to improve their recommender system. For this purpose, the company released a large dataset of movie ratings. This competition was notable in a few ways. First, the prize pool was one million dollars, and that was one of the main reasons it became famous. Second, because of the prize, and because of the dataset itself, many researchers invested their time into this problem and that significantly advanced the state of the art in recommender systems.
It was the Netflix competition that showed that recommenders based on matrix factorization are very powerful, can scale to a large number of training examples, ...
