The Research And Application Of Personalized Recommendation Based On The Collaborative Filtering Of User Clustering

The rapid development of Internet brings not only a wealth of information,but also some problems such as information overload,information waste at the same time.First,the information overloading problem seriously affects consumers.It is difficult for consumers to digging their real interest from a large amount of information.Second,information overload problem has plagued by serious business.Currently,we are facing with consumer upgrades.This requires businesses to the original scale management strategy into differential operation strategy,in terms of each custom's difference to provide the personalized service.Therefore,a more active,more personalized recommendation platform,for consumers and businesses are actually needed.collaborative filtering technology is used most widely in the personalized recommendation system.But with the deepening of the application of collaborative filtering in personalized recommendation,it also exposed many problems.This article mainly done the fol owing research works:1.This paper first introduces the development process of recommender system,then introduced the algorithm theoretical basis which are widely used in the current recommendation systems.Mainly includes the recommendation based on association rules and the collaborative filtering two kinds big.Points out the problem that existed in the col aborative filtering recommendation algorithm---"sparse data";2.For "sparse data" problem in collaborative filtering algorithm,this paper proposes a collaborative filtering algorithm based on user clustering.This clustering method in view of the original k-means clustering algorithm in the center of the selection of initial value point,through calculate the difference between the data to redefine the selection principle of initial center,it had avoided the uncertainty of the clustering results due to the random initial center and;3.Finally,For the improved collaborative filtering algorithms validation,using the proposed improved recommendation system into “douban users books recommended" experiment.And the experiment shows the detailed steps about how to use the improved recommendation system for users to recommend.