A Complementary Predictor for Collaborative Filtering

Recommender systems provide users with personalized suggestions for products or services. Collaborative Filtering (CF) is a method commonly used in recommender systems to establish a connection between users and products by analyzing past transactions. There are two successful approaches to CF, one is the Slope One scheme, which is efficient to query and easy to implement, and the other is the latent factor models, which directly profile both users and products. In order to obtain better accuracy and efficiency, a new combined approach is proposed by taking advantages of both Slope One and SVD models. The method is tested on the MovieLens dataset with the experimental results showing that the proposed final solution may achieve great improvement in prediction accuracy when compared to using the Slope One or SVD models alone.