Personalized fitting recommendation based on support vector regression

摘要: Collaborative filtering (CF) is a popular method for the personalized recommendation. Almost all of existing CF methods rely only on rating data while ignoring some important implicit information in non-rating properties users and items, which has significant impact preference. In this study, considering that average items certain stability, we firstly propose fitting pattern to predict missing ratings based similarity score set, combines both user-based item-based CF. order further reduce prediction error, use attributes, such as user’s age, gender occupation, an item’s release date price. Moreover, present deviation adjustment support vector regression. Experimental results MovieLens dataset show our proposed algorithms can increase accuracy recommendation versus traditional