Analytics Team

Check out this Techcrunch article highlighting Myspace’s new recommendation system, Qizmt. It’s built on a mapreduce framework.

In an effort to increase user engagement, satisfaction, and profitability, many websites are offering their users various types of recommendations. Many people are familiar with Amazon.com’s people who bought also bought or people who browsed also browsed. But how do they do that?
Darren Vengroff, chief scientist from RichRelevance, explains some of the components [...]

Netflix just implemented some updates to their star rating predictions based on some work done on the Netflix Prize. Here’s what Todd Yellin, Director of Product Management, and Jon Sanders, Director of Recommendation Systems have to say on the Netflix Blog.

Greg Linden put together a few thoughts about recommendation algorithms. His ideas continue at a post on the Communications of the ACM Blog.

The NY Times highlights the problem that recommendation systems like Netflix sometimes have. For movies like Napoleon Dynamite, people either love them or hate them and it is very tricky to predict which it will be.