Research On Collaborative Filtering Recommendation Algorithm Based On User Attribute Clustering

As a classic personalized recommendation technology,collaborative filtering has been successfully applied in many fields and has made great progress.With the rapid development of cloud computing,Internet of things technology,our information processing ability cannot catch up with the growth rate of information.Information overload becomes increasingly prominent,especially under the circumstance when the amount of data in the future will achieve the ZB,collaborative filtering recommendation technology with three problems: scalability,cold start,and data sparseness will have more and more severe challenges.For the scalability and cold-start problem faced by collaborative filtering recommendation algorithm,the similarity model is optimized,and the PCEDS similarity model based on optimized K-means method is proposed to improve the recommendation quality.With the rapid development of clustering technology,the clustering method is becoming more and more mature,and K-means clustering method is the most classical and easy to implement clustering algorithm.Aiming at the scalability problem of collaborative filtering recommendation algorithm,this paper proposes a collaborative filtering recommendation algorithm on the basis of user attributes characteristic information of K-means clustering collaborative filtering recommendation algorithm,because the K-means method of the selection is sensitive to the initial clustering center,in this paper,the K means clustering is optimized,which means selecting the initial cluster centers based on user activity(CF-act).Before the optimization,this paper firstly analyzes and arranges the user attribute feature information.In order to solve the cold start problem of collaborative filtering recommendation algorithm,this paper proposes a collaborative filtering recommendation algorithm that fuses user attribute feature similarity and user preference similarity.Due to the data processing of user attribute feature information,this paper proposes a trust-based user attribute similarity model,which combines the user attribute features based on trust and the user preferences similarity model and that proposes a novel PCEDS similarity model establishes a predictive model of clustering results.The experiment result illustrates that the proposed PCEDS algorithm can significantly reduce the root mean square(RMSE)compared with the traditional collaborative filtering recommendation algorithm,and significantly improve the recommendation precision and recall,which can ease cold start problem,and simultaneously clustering technology can reduce running space of system memory and reduce recommendation time.