Top-N Recommendation Learning Problem

Definition

Top- $N$ Recommendation Learning Problem

The Top-$N$ Recommendation Learning Problem is the task of learning, from observed user–item data, a recommender that ranks candidate items according to their predicted relevance to a user.

Given a user $u$ and a set of candidate items $I$, the learned model assigns each item $i\in I$ a relevance score

$$r(u,i),$$

and recommends the $N$ highest-ranked items.