Movielens is a personalized movie recommendation site run by the Grouplens research team at the University of Minnesota. I've known about the site for a while but only recently started using it. So far I've been impressed.
Ideally though, I would like the following technology to exist. My girlfriend and I walk into Blockbuster and assume our regular routine of scouring the DVD racks looking for movies that we remember we should see. We might pick up a DVD, read the back of the movie box and keep it in our hands just in case we decide on it. I would like to be able to use my cell phone to take a picture of the front cover (or perhaps the DVD bar code) and query Movielens right there in Blockbuster. Movielens would then automatically take into account that I am with my girlfriend (because we both have bluetooth enabled cell phones) and tell me whether this is a movie we would both enjoy (e.g. "Yes, based on your mutual preferences, Hotel Rwanda is a 4.5"). Movielens could also return the imdb.com ranking along with the rottentomatoes.com ranking to provide a bit more context to their personalized rating.
Update: [9:42 05/04/2005] So, since I made this post (3 long days ago), I found three papers that nearly address my above scenario: PocketLens - Toward a Personal Recommender System by Bradley Miller, et. al., MovieLens Unplugged - Experiences with a Recommender System on Four Mobile Devices by Bradley Miller, et. al., PolyLens: A Recommender System for Groups of Users by Mark O'Connor, et. al. All of these papers can be found at the University of Minnesota Grouplens site.
Sunday, May 01, 2005
Subscribe to:
Post Comments (Atom)
No comments:
Post a Comment