Design And Implementation Of Recommender System Based On Clustering Algorithm

The rapid development of the Internet has brought great convenience to people at the same time,there are also some problems,the so-called "information overload",in the era of information explosion,information has become increasingly rich,how to find the information they want from the vast information has become a hot research topic in the film,as one of the people entertainment,there are the difficult choice of disease,in order to solve this problem,the personalized recommendation system.The algorithm recommended the most popular system for collaborative filtering,the algorithm is simple and easy to understand,but in the actual use,in the face of the sparsity of user data,decrease recommendation quality has become a problem,this paper analyze the current mainstream recommendation algorithm,and summarizing its advantages and disadvantages,taking into account the characteristics of the film score itself,using non negative matrix factorization algorithm to preprocess the data,proposed an improved collaborative filtering algorithm.The main research work of this paper is as follows:(1)firstly,it introduces the background and the domestic and foreign research situation and the development of recommendation system in recent years.(2)analysis of some of the current mainstream algorithm of collaborative filtering algorithms such as simple and classic SVD++ algorithm,SVD algorithm and matrix decomposition has the KDD shine Cup match and mixed to comprehensive advantages of each algorithm recommendation algorithm based on the model,the emphasis of the above algorithm,summarizes the advantages of various algorithms and problems the problem,in order to pave the way for the later part of the.(3)for now the algorithm's shortcomings,combined with the film score of negative score is of no significance,this paper proposes a recommendation algorithm based on non negative matrix factorization,the idea is first to the reality of the existence of the sparse data set for non negative matrix factorization,the purpose is to reduce the data sparseness and dimensionality reduction,and then nearest neighbor clustering recommendation.(4)compare the performance of the improved algorithm and the traditional algorithm on the Movielens data set,and study the influence of the parameters on the algorithm,and verify the effectiveness of the improved algorithm compared with the traditional algorithm.(5)based on the improved algorithm,a prototype system is implemented,which is based on the requirement analysis of the system,the design of the database and the realization of the prototype system.