In the paper "Item-Based Collaborative Filtering Recommendation Algorithms" In section 3.2.2 about regression, it is said the the user's actual rating of item N (Ru,n), is replaced with an estimate of that rating (R'u,n) via linear regression. The linear regression model seems to be mapping rating of the target item (i) to ratings of the similar item which the user has also rated (N). Question is, what is input into the linear regression model for the estimate (R'u,n) to be calculated. It can not be the target user's rating of the target item (Ru,i) since that is what we are trying to predict in the first place. Is the input into the regression the target user's rating of the similar item (Ru,n)?