A Study on Collaborative Filtering Recommendation Algorithms

Abstract

Recommendation system has proved to be an effective tool to solve the problem of information overload. In particular, collaborative filtering (CF) is considered to be the most popular and widely implemented technique for recommendation systems. In this paper, the conventional CF algorithms are first introduced in terms of basic concepts, principles, and evaluation indicators, and the issues related to data sparsity, scalability, and cold start are analyzed. Then the available findings and solutions of these issues are summarized. Finally, the research hotspots of CF recommendation algorithms in interpretability, data privacy, recommendation based on deep learning, and combination with cloud computing big data are put forward. This paper presents a complete framework of CF knowledge, which will clarify the research context of CF and provide reference for follow-up studies. This work will be of certain significance in advancing the progress of personalized information services.