Comment by kozikow

I am curious why you used the user-closeness model.

Matrix factorisation won the Netflix prize: http://dl.acm.org/citation.cfm?id=1608614 . What made you go for user closeness?