Research On Collaborative Filltering Technique In The Application Of Resources' Personalized Recommendation

With the rapid development of e-commerce, the information of the network resources is also constantly expanding. People enjoy the convenience of the network, at the same time, also under the pressure of large amount of simultaneous presentation of information. In order to improve the network information service more effectively, personalized recommendation system came into being. Collaborative filtering recommendation as the current mainstream technology has been widespread concerned and studied by people, but it still remain some hot issues, such as the problems of sparseness, cold start and scalability etc, to be resolved. However, there are personalized recommendation technology is not yet mature, and still faces a series of challenges.In this paper, we make a deep research on the personalized recommendation systems and related technologies. According to the problems of sparsity and scalability that algorithm exists, we analysis the improved methods the existing, and proposed an algorithm set of methods on this basis, while addressing these two issues. In order to reduce the score set of data sparseness, this paper uses the method of Slope One to fill them. Facing the high-dimensional rating data which has been filled, this paper introduces the PCA-SOM technique. Through the PCA dimension reduction techniques to ensure the score data of the main information, then introduces the SOM clustering to narrow the scope of the nearest neighbor search. The whole process of the improved algorithm uses off-line model, and reduces the time complexity of online recommendation, so effectively improve the system scalability issue. Compared the improved algorithm with the traditional algorithm in the recommendation accuracy, shows that the improve algorithm has more accuracy in recommendation, and the experiment set based on the standard data. The result also shows that the improved algorithm in a certain extent, improved the quality of the recommendation system.Finally, through the requirements analysis on the teaching resources website for personalized recommendation service, in this paper, design a personalized teaching resources recommendation system. According to the characteristics of online teaching resources system, discussed the main functional modules of recommendations. We apply the improved collaborative filtering technology to the teaching resources platform, effectively meet users'needs which for personalized recommendation service.