December 14th, 2005


Recommendation: MovieLens

I know several of my friends love rating things. They spent hours going through their music collection on iTunes, deciding if a song is a 3 or a 3.5, putting things into one playlist or another. To those people, I invite you to try MovieLens, a University of Minnesota project in collaborative rating that I've been using for several years. It's simple - you get lists of movies, you pick a rating using pull-down menus, and the system predicts what other films you'd like based on what you've seen and how you've rated them. I'm up to 867 rated films in their system right now, and it is scarily accurate. Usually, their prediction is within half-a-star of what I think about the film upon seeing it.

If you join up, you can add me to your buddy list using the email address seen on my info page. That lets you get joint recommendations where you see both your predicted rating and and your friend's rating.

Oh, and here's where statistics prove interesting -- it tells me some things about my ratings in comparison to the rest of their user base.

Collapse )