Jinni Movie Recommendation System Not Quite Out of the Bottle

"Jinni says it is building a better movie recommendation system by filing films under a long list of granular categories, such as plot, time period, mood, and audience. According to Spanier, the company uses a dual approach of automated semantic metadata processing of things like existing movie reviews and film professionals to provide further insight. Jinni has indexed over 15,000 titles so far, focusing not on a particular genre or just new releases, but on "titles people would rent on DVD or buy online.""

Recommendation engines are big business and something movie rental companies are certainly interested, hence the existence of the Netflix prize.  Looks like this on is at the invite-only beta stage, so if you've taken a looks, let us know what you think.