OPINION: Standing out
By Paul Sweeting -- Video Business, 12/12/2008
|
|
Recommendation engines have long been a feature of online retail sites, from Amazon.com to Netflix to iTunes. Yet for the most part, they’ve been regarded as a tool for customer retention and improving the user experience. Rarely have they been the focus of a competitive arms race themselves.
That may be changing, however.
Last week, Netflix awarded its second annual Progress Prize of $50,000 to a group of seven computer scientists and machine-learning experts working under the name “BellKor in BigChaos,” for improving Netflix’s recommendation engine by 9.44%.
The group, many of whom had competed in the Netflix challenge last year, improved on its previous improvement performance of 8.43%, but once again missed the grand prize of $1 million, to be awarded to anyone who can improve the recommendation engine by 10%.
Netflix didn’t say what criteria it applies in measuring “improvement,” only that a “board of judges” validated BellKor’s work. But if you can miss out on a million bucks by 0.66%, frankly, the whole deal sounds rigged to me.
Read the full column on ContentAgenda.com.
