Group Comes Close to Winning $1M Netflix Prize

NEW YORK -- A multinational group of researchers, scientists and engineers are close to winning a $1 million challenge to improve Netflix's system of recommending movies that its subscribers might like.

The online movie rental company had launched the Netflix Prize contest in 2006 to improve its predictions by at least 10%. The idea was to farm out valuable research to thousands of enthusiastic participants.

A team called BellKor's Pragmatic Chaos submitted its solution last week, saying it has improved the predictions for what movies people will enjoy by 10.05%.

But the group must wait before being declared winners. The team's solution kicked off a 30-day period during which other contestants can enter their best work and possibly beat the BellKor team's threshold. Netflix spokesman Steve Swasey said, however, that the company had yet to receive any other submission breaking the 10% threshold.

There are currently 49,430 participants from 184 countries taking part in what Swasey called a "fierce global competition."

While the prize Netflix has dangled clearly has attracted a lot of attention, as the contest has advanced and more teams collaborate, the money will have to be split into more chunks. If this team wins, each of the seven members will get just $142,857.14.

BellKor's Pragmatic Chaos consists of four teams that have decided to collaborate for the prize — two researchers at AT&T, two engineers from Montreal, a research scientist at Yahoo and two machine-learning researchers from Austria.

Chris Volinsky, director of statistics research and a member of the team, said Netflix has presented a really interesting analytic problem that's couched in something everyone can relate to: movies.

And, he adds, it's not about the money, rather, it's an academic pursuit that's also been "a lot of fun" in the competitive sense.

"For geeks, this is like watching the World Series, the Super Bowl, the Indy 500 and the Stanley Cup all at once," Swasey said.