Tuesday, October 03, 2006

NetFlix Offers $1 Million to Improve Movie Recommendations..

NetFlix is offering a prize of $1 million to anyone who can improve its current movie recommendation system by at least 10%. To facilitate this, the company is is publically making available 100 million of its customers’ movie ratings.

NetFlix may have released the largest database of its kind to the public, however it is insufficient for creating an effective movie recommendation system solely on this data. This data becomes insufficient for 2 primary reasons:


  • Customers rent movies for members of the household. A mother may rent movies for both her child and herself. Therefore if a mother rents 8 movies a months (6 educational documentaries for her child and 2 romantic comedies for herself), does this mean she likes documentaries?

  • People share NetFlix accounts. Along the same vain as my previous point, many households share a single account. Therefore if my wife enjoys comedies and I like suspense films it will not be fair to conclude that a single account holder enjoys both comedies and suspense flicks.


The idea of using prize money to attract programmers is not new. Google offers an annual competition where it gives away money. A discovery may lead to broader applications. As one startup CEO in a Red Herring article quotes “Intelligent systems that know how to recommend things just for you are the next generation of search.”

If you are interested in going for the $1 million by NetFlix, here are the details.

No comments: