Collaborative Filtering Recommendation Algorithm Research Based On Combinatorial Similarity And User Feature Clustering

Since the 21 th century,internet technology and e-commerce develops rapidly,all kinds of network data had shown a exponential growth trend,so the users were unable to quickly obtain useful parts for themselves because of the excessive information.In view of the above problem of information overload,the recommendation system came into being and was used to alleviate the above situation.At present,collaborative filtering recommendation algorithm is widely used among many personalized recommendation technologies.However,with the increase of the number of users and the number of items,the scale of the system continues to expand.The traditional collaborative filtering recommendation algorithm still faces many issuses,such as data sparsity problem,cold-start problem,scalability problem,and real-time problem.On account of the above problems,this paper proposed an improved collaborative filtering recommendation based on combinatorial similarity model and user feature clustering.First,a newly users similarities calculation model,integrating the improved user ratings similarity and item category preferences similarity,was proposed;then,considering the K-means clustering algorithm,we clustered the users based on the user attributes through Euclidean distance,so as to cluster the target users in the properly clusters.Finally,the target users could get the score prediction and items recommendation.In order to verify the superiority of the improved algorithm,the improved algorithm is compared with other improved collaborative filtering recommendation algorithms on MovieLens data sets.The experimental results show that the improved algorithm not only efficiently alleviates the problem of similarity measurement imprecision caused by the extreme sparsity of user rating data but also effectively improves the recommendation instantaneity and expansibility.