Collaborative Filtering Recommendation Algorithm Based On K-SVD

With the rapid development of the Internet,network information also showed exponential growth,information overloading has become a daily problem.The collaborative recommendation system based on collaborative filtering is a great way to solve the information overload problem,which can make people get more accurate information what they want.The reason why collaborative filtering recommendation algorithm has been able to achieve great success is it uses user's scoring matrix to calculate the users of similar users,which finds similar users more accurate than the traditional content-based recommendation.However,the collaborative filtering algorithm has some limitations because the user scoring matrix is a sparse matrix.How to use the sparse matrix to calculate the similar users of the specified users and then complete the recommendation of users become the main problem of cooperative filtering recommendation algorithm.In this paper we will first introduce the background of the recommendation systems and the core recommendation algorithms of the mainstream.Comparing the traditional recommendation algorithm and the cooperative filtering recommendation algorithm,we can see the advantages of the cooperative filtering recommendation algorithm.Secondly,we deal with the problem of sparse matrix in collaborative filtering recommendation algorithms by using statistical filling,machine learning filling and SVD decomposition.It does not play a good effect in the criteria of recommendation time and accuracy.Then,in this paper we proposes a collaborative filtering recommendation algorithm based on k-SVD.The recommendation algorithm uses the user characteristic matrix to calculate the similarity between users,and uses the feature matrix to calculate the similarity between the films.Then,the cooperative filtering algorithm Complete the recommendation.Comparing with the cooperative filtering algorithm based on machine learning,Collaborative filtering recommendation algorithm based on k-SVD reduce the deviation of the sparse matrix.Comparing with the cooperative filtering algorithm based on k-SVD preserves more information than the cooperative filtering algorithm based on machine learning.Finally,comparing with different Collaborative filtering recommendation algorithm,the effect of the cooperative filtering algorithm based on k-SVD are significant from the time and accuracy.