I am working on github contest with Daniel Haran. Unlike NetflixPrize, the github contest is a Top-K resys task. I think, it is another important task in recommender system.
Let's take movie recommender system for example. When we design a movie resys, we meet two problems:
1) Given a user, we should find which movies he/she will watch. That is finding a candidate movies set.
2) In the candidate set, we should find which movies this user will like after watching.
I think, the first task is top-k recommendation task (GitHub) and the second task is prediction task (NetflixPrize).
Solving two above tasks is the fundanmental of design good recommender system.
The Prize of github contest is A bottle of Pappy Van Winkle 20 year reserve - the best damn bourbon whiskey ever (age permitting) and a free large GitHub account for life.