The $50,000 algorithm

Before jumping into the great work by Donald Knuth, I thought it would be interesting to start with analysis of a contemporary algorithm which has already given its authors $50,000.

NetFlix currently has a $1 million contest to anyone who can write an algorithm for predicting user ratings that beats their algorithm by at least 10%. This content has run since 2006 and so far, no one has done it.

Each year, if no one has claimed the top prize, NetFlix awards a $50,000 progress prize to the best algorithm to date. The first such prize was given this year. In order to claim the prize, the individual or team that wrote the algorithm must publish its details.

This year's prize went to Robert M. Bell, Yehuda Koren, and Chris Volinsky, members of the AT&T Research Labs. Considering that there are over 9,000 teams competing for the prize, AT&T can be very proud of their achievement.

As a starter to this blog, I thought it would be interesting to go in detail through the algorithm that won the 2007 Progress Prize.

References

• NetFlix Prize, Home Page
• "Netflix Prize", Wikipedia
• Robert M. Bell, Yehuda Koren, Chris Volinsky, "The BellKor solution to the NetFlix Prize"