The crosses are a nice idea, although we'll be hard at work at improving our algorithm first (one guy in the team is fulltime on that) and see how it goes.
We'll try to get as much as we can right by default until we ask for help from users :)
I would be curious to know what kind of algorithms you would be looking at. I am a researcher with research interests in "text analytics" and "social media" & if works well I would be glad to contribute.
We're currently working on an algorithm based on multiple criteria like filesize, quality, release group...
Our first goal is to find the best torrent for a chosen movie.
Maybe we'll use social medias for movie matching.
We'll try to get as much as we can right by default until we ask for help from users :)