Recommendation Model Based on Collaborative Filtering Recommendation Algorithm

Abstract

There are problems concern the current recommendation model such as the information recommended is not inaccurate enough.This paper presents a collaborative filtering algorithm based on Kmeans algorithm.Firstly, we analyzed the similarity calculation method of collaborative filtering recommendation algorithm, then we proposed a valuation formula based on user rating scale and information popularity to assign value for ungraded items at sparse ratings matrices to improve the scoring matrix density, increase the accuracy of similarity calculation, and build the recommendation model.Simulation results show that the proposed collaborative filtering recommendation algorithm based on K-means has higher prediction accuracy and classification accuracy than traditional collaborative filtering algorithm.