Nowadays, item-based Collaborative Filtering(CF) has been widely used as an effective way to help peoplecope with information overload. It computes the item-itemsimilarities/differentials and then selects the most similaritems for prediction. A weakness of current typical itembasedCF approaches is that all users have the same weightin computing the item relationships. In order to improve therecommendation quality, we incorporate users’ weightsbased on a relationship model of users into item similaritiesand differentials computing. In this paper, a model of userrelationship, a method for computing users’ weights, andweight-based item-item similarities/differentials computingapproaches are proposed for item-based CFrecommendations. Finally, we experimentally evaluate ourapproach for recommendation and compare it to typicalitem-based CF approaches based on Adjusted Cosine andSlope One. The experiments show that our approaches canimprove the recommendation results of them.
展开▼