Research Of Weighted Slope One Algorithm Based On Clustering

Recommendation system is a product of the rapid development of the Internet,and it has been become more and more important in our daily lives,work and study.Nowadays,the recommendation system has been developed rapidly in the fields of e-commerce,film,social,and so on,and the application and research of the recommendation system had always the research focus at home and broad.The core of the recommendation system is the recommended algorithm and the big data it depends on.The most commonly used recommendation algorithm in recommendation systems is Collaborative Filtering(CF).In this paper,the more classic Slope One algorithm in collaborative filtering algorithm is studied in depth,and the following two improvements are proposed according to the defects of the algorithm:First,in view of the original Slope One algorithm does not consider user similarity and data sparse cases,the recommended effect is not as good as the traditional collaborative filtering algorithm and other issues.This paper proposes a cluster-based user-weighted Slope One algorithm.First of all,to solve the problem of scoring matrix sparsity,this algorithm introduces the singular value decomposition technique to reduce the dimension of the scoring matrix.And then,the improved k-means algorithm is used to classify the scoring matrices after the dimensionality reduction filling,so as to reduce the search time of the neighbors and improve the scalability of the algorithm.The last fusion of user similarity to improve the traditional Slope One algorithm does not consider the issue of user similarity.Second,The original Slope One algorithm does not consider the project similarity and the original similarity measure can not mine the internal characteristics of the project itself.In addition,with the passage of time,the user interest will also be changed,so that the accuracy of the algorithm may be deviated.In order to solve these problems,this paper presents a clustering-based project-weighted Slope One algorithm.In order to merge the similarity of the original Slope One algorithm and mine the categories of the project itself,the algorithm first classifies the project according to the characteristic attributes of the project using the clustering algorithm or the natural category of the project itself,The project is divided into a number of categories,so that you can calculate the similarity of the category of the targetproject,to a certain extent,make up for the problem caused by data sparse,and then in order to highlight the role of new data to weaken the old data on the score impact,the introduction Time weighting function.Finally,the item similarity of the same item in the target item and the K-nearest neighbor are obtained and integrated with the time weighting function into the weighted Slope One algorithm for the score prediction.In order to validate the two algorithms proposed in this paper,we use the real MovieLens data set as the experimental data source,and compare the two proposed algorithms with other existing algorithms respectively to verify the proposed algorithm.Experimental results show that the proposed algorithm can effectively alleviate the sparsity problem of scoring matrix,reduce the search time of nearest neighbor users or projects,improve the accuracy of scoring prediction,and have better scalability.