The Research Of Collaborative Filtering Recommendation Algorithm Based On The Fuzzy C-means Clustering

With the rapid development of modern information technology,people from the lack of information era have entered into the era of information overload.To some extent,it will take people more time in obtaining valuable or interesting information,but the recommendation system has the better performance in solving such problems.According to the user interest preferences,the system can generate the personalized recommendation and effectively solve the negative effects of information overload.The core of recommendation system is recommendation algorithm.Among various kinds of recommendation algorithms,collaborative filtering recommendation algorithm is the most widely used.With the rapid growth of data scale,the problems of collaborative filtering recommendation algorithm are more and more obvious,mainly including data sparsity,cold start,scalability,and the change of user interest over time.About the existing problems,this paper introduces the fuzzy clustering algorithm and proposes a new collaborative filtering recommendation algorithm based on fuzzy c-means clustering and the time weighting function.About data sparsity,fuzzy clustering is carried out on the project attribute information in order to make projects belonging to different categories in different membership degrees.The method increases the data density,effectively reducing the data sparsity,and solves the issue that the inaccuracy of similarity leads to the low recommendation accuracy.Before fuzzy clustering,subtractive clustering is used to determine the optimal initial clustering center as the input of FCM algorithm.About the issue of user interest migration,this paper introduces the time weighting function.The scores evaluated at different time are given different weight values,and recently evaluated projects are more representative of the user current interest,so we will give higher weight values,and early evaluated projects have less effect on the user current interest,thus the weight values are relatively lower.Finally,this paper makes an empirical analysis on the MovieLens dataset,and determines the optimal parameters about the improved algorithm,and validates the effectiveness of the initial clustering center based on subtractive clustering and the effectiveness of the time weighting function.As the same time,in order to illustrate the superiority of the improved algorithm,it is compared with the traditional collaborative filtering algorithm.