The Recommendation Technology Research Based On User Attribute Clustering And Svd Algorithm

With the increasing popularity of the internet, E-commerce has already turned to be a kind of fashion. More and more users on the internet join into different kinds of E-commerce websites. Then, an enormous advanced development space is left for the recommendation system. With the expansion of the system, the most important goal for every recommendation system is how to make the personalized recommendation for different customers rapidly and effectively. Collaborative filtering recommendation is one of the most popular technologies among the E-commerce system, but it causes some problems, such as "cold start" and "data sparse". Meanwhile, most of the recommendation system ignores the features of the user, such as age, gender and occupation when they make the recommendation and it results in the inaccurate recommendation. In conclusion, the main purpose of the paper is how to improve the accuracy of recommendation by considering the features of the user under the condition of sparse data.In this paper, the problem of data sparse is mainly studied. On the basis of deeply analysis of various common recommendation algorithms, we propose a hybrid recommendation algorithm combining singular value decomposition (SVD) algorithm with a special clustering method. The algorithm can improve the accuracy of recommendation in the case of sparse data. The main tasks are as follows:1. Introducing the user-based and item-based collaborative filtering techniques, then we discuss the advantage and limitation about this two techniques.2. Introducing the methods of clustering and analyzing the importance of user attributes. Put forward a new approach to calculate the characteristic value of users, and clustering users in the system with K-means method.3. Introducing SVD method, and put the improved SVD method into the recommendation system.4. Taking advantage of the algorithm approached in the paper, we simulate a recommendation system engine and recommendation items to the target user. Through the engine, we show the recommendation procedure of the algorithm.5. Design some experiment to compare the recommendation quality among different recommendation algorithm. Finally, we compare and analyzing the experiment result data which the SVD based clustering method and other traditional method on the same publicly available dataset and analyze the results.