If the user Ui is active and has sufficient ratings in the training set, the item Vj is frequently accessed in the training set, then we will have accurate estimation of user factor Ui and item factor Vj . In this case, rating prediction will be the expectation of the generated rating from collaborative filtering and expert opinions.