September 22, 2009, 8:14 AM — In 2006, Netflix announced a contest: they'd give $1 million to anyone who could improve their movie recommendation engine by 10% or more. On Monday, September 21st 2009, that prize was awarded to a team that call themselves "BellKor's Pragmatic Chaos." The team, lead by two members of AT&T's Statistics Research Department, met face-to-face for the first time when they attended a news conference to accept their prize. The team's members come from the US, Austria, Canada and Israel.
Their winning entry was measured at 10.06% (according to the LA Times) better than the existing Netflix recommendation engine. More specifically Netflix was measuring how accurately the submitted algorithm could predict user scores for a given movie, based on that user's ratings for other movies.
The winner was actually determined back in July. After almost 3 years of competition, Bellkor's Pragmatic Chaos edged out "The Ensemble," by submitting first. A story at CNET claims their scores were identical but Bellkor's Pragmatic Chaos submitted 24 minutes earlier than The Ensemble and so were awarded the price. (According to their website, The Ensemble submitted their final attempt with just 4 minutes left until the deadline — they weren't going down without a fight!)
Wish you'd gotten in on the fun and the chance to win some big money? It isn't too late, because Netflix wasted no time in announcing a second contest. Netflix Prize 2 asks teams to come up with a recommendation engine that bases its predictions on user demographics and prior rentals. In other words, a way to make recommendations for users who haven't bothered to rate any movies. This next prize will be awarded in two parts: $500,000 to the best entry after 6 months and another $500,00 to the best entry after 18 months. The new contest hasn't officially begun yet, but keep an eye on www.netflixprize.com for details.
And congrats to Bellkor's Pragmatic Chaos!
Follow Peter on Google+
