The Improvement Of Collaborative Filtering Recommendation Algorithm

With the popularity of the Internet and the rapid development of information technology, the quantity of the Internet users and the amount of the information is also growing rapidly. It has became more and more difficult for people to obtain quickly their interested information from the vast amounts of information. User-centered site such as social media, e-commerce, etc, information is more and more enormouse, producing the vast amounts of data that related with user interest, however, the information of the user's attention just a few. Personalized recommendation technology emphasizes embarks from the user's interests, according to different users with specific personalized recommendation service. Personalized recommendation technology does not require users to provide specific requirements, but staring from the historical behavior and data of the user, on the basis, establishing a relevant model for mining the user's interests and needs, thus selecting the information which users interested from the vast amounts of information. Therefore, when personalized recommendation technology is not clear forusers'requirements, it is particularly important.So far, many recommendation algorithms have been put forward, but collaborative filtering recommendation algorithm is the most widely used in the recommendation algorithm and one of the most successful algorithm.However, although the collaborative filtering recommendation algorithm has been successfully applied to many recommendation systems, with the expansion of the system,the number of users in the system and project quantity is increasing, collaborative filtering recommendation algorithm are faced many severe challenges suchas data aparseness, cold start problems and scalability problems, etc.Aiming at the problems of the collaborative filtering recommendation algorithm, this paper proposes a combination of collaborative filtering recommendation algorithm. First of all, this paper expounds the basic idea of collaborative filtering recommendation algorithm, and expounds some problems existing in the algorithm. Then in view of the problems put forward an improved algorithm of clustering based on SVD and hierarchical clustering, and an improved algorithm of the Slope One. In order to verify the validity of the improved algorithm, contrasting experiment on MovieLens data set and analyzing results, verifing the validity of the improved algorithm. The experimental results show that the recommendation quality of theimproved algorithm is better than traditional recommendation algorithm. In the end of this paper, we propose a combination of the two improved algorithms in combination of collaborative filtering recommendation algorithm. Experiments are performed on the MovieLens data set, and the experimental results verify the effectiveness of the combination of collaborative filtering recommendation algorithm.