how evaluate the top-N recommendations?

[huda]

HI
I'm Phd student, working in Recsys \collaborative filtering.
I'm still new in this field.
I have query.

I know how prediction ratings are evaluated , but I don not know how recommendation lists are evaluated.
My work is nor relate with ratings and predicate the ratings but with recommendations list.

if I predicate an list for one user, how I know if this list is right or not.
where this list has new items for user that he is not know it , buy it, read it, show it, etc.

I have problem in understanding this issue.

thanks in advance