Personalized Recommendation Research Based On Collaborative Filtering And Improved Algorithm

With the rapid development of Internet,the data information has grown rapidly.On the one hand,the emergence of vast amounts of information has made it difficult for users to find the interesting content.On the other hand,it has made large amounts of data less interesting.In order to solve the problem between the users' needs and the huge amount of data on the Internet.Personalized recommendations have attracted more and more attention in the academic and industrial fields.Due to the dramatic increase in the number of users and the number of projects,the recommendation system has a serious data sparsity problem.This article comprehensively analyzes the research status at home and abroad,and it makes an in-depth study on how to improve the accuracy of personalized recommendation system.First of all,this article introduces several commonly used other personalized recommendation technologies,and analyzes and compares the advantages and disadvantages of each technology.Then,the process of collaborative filtering algorithm is deeply studied.The sparsity and cold-start problem of CF algorithm are analyzed and studied in detail.A solution to singular value decomposition is proposed for sparseness problem.The value is filled,and then SVD is used to reduce the dimension of the filled matrix.Finally,a project attribute and penalty factor are introduced to improve the similarity calculation formula.By modifying the similarity of the original project,the similarity measure of the traditional project is improved.In order to get the target user's nearest neighbor collection,finally get the target user's unknown project rating prediction.By improving the dimension reduction and similarity algorithm,the accuracy of the recommendation system is improved.In addition,the user's interest is not static and will change over time.Therefore,from the perspective of the time dimension,the time-weighting of the user's interest and the time weighting of the items are performed to calculate similarity,thereby improving the accuracy of the recommendation effect.Finally,the modified algorithm is tested on MovieLens dataset.The results show that the improved CF algorithm has higher accuracy than the traditional CF algorithm.