I've a friend who's going to work for Comcast. He's going to be creating algorhythms to figure out how to recommend movies a lot better. The way he has it, he's doing some advanced geekery, which is right, but i like to think he's going head-to-head with netflix's rating system (which i think is par).
I told him that he needs to build in some kind of device to create explicit associations between movies. For instance, i saw the Illusionist and The Prestige. Both wer movies about magicians. However, there's nothing except word of mouth to say that these movies are alike. For all of that, if i searched for "magic", i would have gotten things like Narnia, LOTR and so forth.
He doesn't even have to make the associations himself! He can have people tag movies and create a little device so that people can vote/say "like" or "not like" other movies along certain categories.
When i was ineffectually telling him this, he said i didn't know how powerful the algorhythms were to identify movies i'd like. So if you or anyone else is intereseted in this, here's a link to an article by the head-honcho of netflix explaining their recommendation system, and how they're trying to open-source thier recommendations improvement.
http://www.forbes.com/free_forbes/2007/0507/130.html
The Best College Majors for a Job that Pays
6 years ago