An improved hybrid recommendation algorithm based on feature preference

Abstract

In order to solve the problem of data matrix sparsity and cold start in the era of big data. This paper designs a collaborative filtering algorithm, which is an improved hybrid recommendation algorithm based on feature preference analysis. It combines the analysis of user feature preference and item feature, and then uses the traditional collaborative filtering idea to recommend the best scoring object to users. The experimental results show that the accuracy of the proposed algorithm is improved obviously.