An Improved Collaborative Filtering Recommendation Algorithm

摘要

The core of the classic collaborative filtering algorithms about similar calculation are designed on the basis of the ȁC;user-item matrixȁD; model. This paper proposes an improved collaborative filtering algorithm on the basis of the ȁC;user-item cubeȁD; 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.