Does it sometimes seem like Netflix knows your taste in movies better than you do? The Financial Post has run a fascinating article that takes a look at the sometimes laughably complex algorithms Netflix uses to recommend movies and genres to its users.
In the article, Netflix’s chief technology officer, Neil Hunt, explains that curating its large catalog of mostly older content is one of the primary tasks of the company. Says Hunt:
“Out of the thousands of titles available there is a subset that is interesting to each individual . . . If we know enough about the content and the individual’s taste we can put those two together and make something that’s more compelling than random choices or the most popular ones.”
Netflix employs a few dozen freelancers who watch and tag videos for the company’s recommendation algorithms. Each piece of content in the company’s library carries more than 100 tags.
Todd Yellin, Netflix’s VP of product innovation, says that mathematics beats humans every time when it comes to predicting customers’ movie tastes. Says Yellin:
“[Hiring movie experts to recommend films] fails miserably next to an algorithmic approach . . . You get your best people to come up with those lists and they don’t do very well. What does much better is a combination of metadata and clustering.”
Netflixing Insiders, is it comforting or a bit disconcerting to know that your movie recommendations are coming from very complex and well-informed computer programs?
[via The Financial Post]