Research Of Collaborative Filtering Recommendation Algorithm Based On K-means Clustering

With the rapid development of Internet and information services,there will be a lot of redundant information in the network.These contents show a trend of exponential growth,resulting in the problem of information overload.The emergence of personalized recommendation system is a good solution to this problem,but also brought great convenience to people.However,due to the increasing size of the system,it will be a lot of problems in the traditional collaborative filtering algorithm.In this paper,we choose to optimize the data sparsity and algorithm scalability.First of all,aiming at the sparsity of data and the scalability of the algorithm,this paper proposes a user clustering recommendation algorithm based on optimizing the clustering center.The algorithm first eliminates the unrated items in the scoring data matrix based on the Weighted Slope One algorithm,and reduces the sparsity of the raw data by preprocessing.Then the K-means algorithm based on minimum variance is used to cluster the pre-processed scoring data.By clustering similar objects together,the nearest neighbor search space of the target user is reduced and the algorithm scalability is improved.Finally,the traditional recommendation algorithm is used to generate the final result.Secondly,aiming at the sparsity of data,this paper proposes a user clustering recommendation algorithm based on improving the similarity calculation.Aiming at the shortcomings of the traditional similarity calculation,the algorithm first corrects the original scoring data matrix based on the time forgetting function to solve the problem that the user's interest decays with time.Then the traditional similarity measures are improved based on user preference and user eigenvector.By introducing some hidden factors,the nearest neighbor search is more accurate and the data sparsity is alleviated.Finally,the traditional recommendation algorithm is used to make recommendations based on the minimum variance K-means clustering.Finally,the user clustering recommendation algorithm based on optimizing the clustering center is implemented on the Movielens dataset.The experimental results show that this algorithm can alleviate the problem of data sparsity and algorithm scalability to a certain extent,and has higher recommendation accuracy.In addition,the user clustering recommendation algorithm based on improving the similarity calculation is also implemented on the Movielens dataset.The experimental results show that this algorithm can deal well with the sparsity of data and improve the accuracy of recommendation.