Movie rental company Netflix confirmed the lucky winners of a $1 million prize, sought for more than three years in a contest to improve the site’s movie recommendation and ratings engine.
In 2006, Netflix introduced a $1 million prize competition that required contestants to improve movie recommendations for subscribers who regularly rate movies they watched. Netflix says 51,051 contestants from 186 countries have tried to come up with a better recommendation system.
Two years into the contest, Netflix started doubting that anyone would come with a solution. But in the end, two teams actually managed to improve Netflix’s predictions engine equally (10.06 percent). The runners-up, Ensemble, lost because they submitted their entry a few minutes later than the winning team, BellKor.
The BellKor team includes two AT&T research staff, Bob Bell and Chris Volinsky; Yahoo’s Israel Lab Yehuda Koren; Austrian researchers Andreas Töscher and Michael Jahrer; and Quebec-based Martin Chabbert and Martin Piotte.
Netflix said that the changes to the recommendation engine are yet to be implemented, as the algorithms contributing to the overall 10 percent improvement are very complex. Netflix hopes the changes will help the company retain subscribers.
Meanwhile, Netflix announced a round two of the company’s prize, which requires contestants to come up with a better recommendation engine for those who don’t rate movies they watched very often. The award consists of $0.5 million to the team that made the most progress by April 2010 and then an additional $0.5 million to the leading team by April 2011.