An Improved Collaborative Filtering Recommendation Algorithm

The core of the classic collaborative filtering algorithms about similar calculation are designed on the basis of the “user-item matrix” model. This paper proposes an improved collaborative filtering algorithm on the basis of the “user-item cube” model, which takes care of the factor of the data produced when the user purchased the item. The algorithm attaches the corresponding weight to the date factor, and then the corresponding weight is used to the calculation of the similarity. This method increases the accuracy of the recommendation system significantly.