Netflix Inc. in October 2006 challenged the world to beat its movie recommendations system by a factor of 10%. Anyone who could was promised the Netflix Prize—$1 million.
A group made up of formerly competing teams under the moniker BellKor’s Pragmatic Chaos was the first to top 10%. Their 10.05% submission set off a ticking clock for all others working on a refined system to beat Chaos. Just before Sunday’s deadline, another team of formerly competing teams, called the Ensemble, claimed on Netflix’s leader board to have beaten Chaos by one one-hundredth of a percentage point.
But that doesn’t mean the Ensemble can claim the $1 million prize. Not yet, at least. Per the rules of the contest, Netflix is analyzing both teams’ refined versions of the company’s Cinematch movie recommendations engine. It will announce the winner in September, the company says.
For Netflix, the point is for its customers to get more enjoyment out of the movies they rent. “The more you enjoy something, the more you tend to do it. And you displace other things you enjoy less,” a Netflix spokesman says. “If today you only find one of every four movies you watch outstanding and we can raise that to three of every four, you’ll want to watch more movies and you’ll enjoy the whole experience more.”
All together, 51,051 contestants on 41,305 teams from 186 countries competed for the Netflix prize. The goal was to substantially improve the accuracy of predictions on how much members are going to like or dislike a movie they have yet to see based on their ratings of movies they’ve already seen and their behavior on the web site. Netflix members rate movies on a scale of one to five stars. The more movies they rate, the easier it becomes for Netflix, No. 18 in the Internet Retailer Top 500 Guide, to guess how members will rate movies they have yet to see.
Such product recommendations engines are becoming a key component on many e-commerce sites. Some e-retailers, like Netflix, have built them in-house. Others purchase systems from vendors such as Baynote Inc., MyBuys Inc. and richrelevance. Many e-retailers have reported significant increases in sales directly attributed to product recommendations.
The winner of the contest will own the rights to their recommendations system. In addition to the $1 million prize, Netflix will pay the winner an annual license fee to use the system.
Back...
