Research On Collaborative Filtering Recommendation Algorithm Based On Clustering

In recent years, with the rapid expansion in sides of E-commerce and the growth of online trading volume, more and more types of goods and mass of information affect the life of customers who are on the Internet everyday. When people are shopping or browsing online during the time in the face of vast amounts of information, they will inevitably become dazed and confused. In order to make users get information and products conveniently, recommender system then generates, it can promote the development of e-commerce rapidly. It also can help users to get more products and information easily which they needed. The role of Recommendation system in e-commerce mall is like the sales staff in the daily life, it provides product information and recommendations to help the customers to find their satisfaction about the goods and complete the purchasing process. Today Recommendation System produces and develops in the context of e-commerce booming, and its application is becoming more and more widespread. The research on the theory by scholars become more and more deeply and widely. This recommendation system for e-commerce makes a more in-depth studying and research.This paper elaborates the utilization and outlook of variety personalized recommendation algorithm in E-commerce application recommendation system base on the analysis of current research, and analyzes the various advantage and weakness. Papers based on collaborative filtering technology to study, and based on the classic collaborative filtering algorithms and present improved algorithms for collaborative filtering methods , a double-clustering based on user and project content and project more effective collaborative filtering recommendation algorithm is Proposed to solving the traditional collaborative filtering algorithms which appears in the data sparseness problem and scalability issues. The thesis of collaborative filtering algorithms improve the school library to borrow personal information online readers to added to the system recommended features and to help users to borrow books more conveniently and efficiently in order to use the library resources effective. Finally Using the experiment to improve the methods and previous research on collaborative filtering algorithm and recommended from the effect of prediction accuracy that were compared, and experimental results obtained, experiments and projects based on the user double-clustering and content integration of multiple projects collaborative filtering algorithms to improve the actual performance of the program.