Netflix Close to Declaring a ‘Recommendation Prize’ Winner30 Jun, 2009 By: Erik Gruenwedel
A worldwide group of computer scientists, software engineers and machine learning researchers is less than four weeks away from possibly winning the top $1 million cash prize offered by Netflix to anyone who can improve the online DVD rental pioneer’s movie recommendation software more than 10%.
Team BellKor’s Pragmatic Chaos, which submitted its answer June 26, must wait a 30-day deadline for additional entries.
BellKor’s submission will then be evaluated by Netflix engineers to determine whether it does improve, by a stated 10.05%, how much someone is going to love a movie based on their movie preferences, according to spokesperson Steve Swasey.
“After the 30-day period, we go into a period of validation to make sure what they submitted is in fact a 10% improvement,” Swasey said. “So we won’t know [if there is a winner] for several weeks.”
According to Swasey, there have been more than 49,000 participants from 184 countries since Netflix launched the competition in October 2006.
During the years several teams have collaborated, including BellKor, which consists of members from Pragmatic Theory and Big Chaos, who separately submitted (less than 10% improvement) solutions this year.
Swasey said the competition has drawn “anybody who is a geek by nature and thinks they can write a good algorithm for collaborative filtering recommendations.”
“If it’s not a 10% improvement, then we will continue,” Swasey said. “If they win the $1 million then the contest comes to a close.”
There have been two, $50,000 progress prizes awarded in the competition, which is slated to end Oct. 2, 2011.