Collaborative Filtering Technology Research And Application Of E-commerce Personalized Recommendation System

With the increasing popularity of networks, e-commerce network has become an important place for merchandise. However, in order to meet the needs of users, the rapid expansion of the volume of merchandise in the e-commerce network, which allows users often get lost in the vastness of the information in the merchandise, it is difficult to find their own satisfaction with the goods. In order to let the customer be able to easily find their own satisfaction goods, it is desperately need to give each customer to arrange for a shoppingleader. E-commerce recommendation system is in such demand for the products.Collaborative filtering technology is the most successful personalized recommendation technology in the field of application. The effect and accuracy of recommendation results show a remarkable advantage. The most widely used Collaborative filtering are user-based collaborative filtering technology and item-based collaborative filtering technology. User-based collaborative filtering recommendation system are very successful in the past, it classify user based on the similarity of user evaluation on items, so the recommendation results are more precise and easier and can easily abtain user's potential interest. But, this recommendation technique has obvious scalability, data sparsity problem, which lead the inaccurate results of user similarty computation, leading to a sharp decline of recommend quality.Item-based collaborative filtering to a large extent solved the data sparse problem and cold start problem. However, the basic of the technique is only on the similarity between items.therefore this technique can only provide the user with the items with which the user is familiar.and the technique can't make a cross-genre recommendation, and is lack of the capacity of serendipity discoverying.In this paper, first of all, we made a deep research on the e-commerce recommendation system, detailed analysis of a variety of personalized Recommend Recommend technology in the field of e-commerce application status and prospects. On this basis, we mainly made research on on the collaborative filtering technology which is used most widely in the area of e-commerce recommendation. Then, made system requirements analysis, system design on a e-commerce system. Made module division on the system and with each of the modules we made a detail analysis of the function requirement. In the part of system implementation, we maily focus on the implementation of the recommendation system. however, the core of recommendation system is recommendation algorithm. Through the research and analysis on the traditional collaborative filtering algorithm, this paper proposed a CIUCF algorithm which is a improved collaborative filtering algorithm, and descibed the idea of the algorithm, the description of the algorithm,the flow of the algorithm.And finally implemented the algorithm,made a detail of the recommendation system architecture,designed the diagram of the class structure of the recommendation system. In the experimental part, first of all, we used the authoritative data sets and took into account all factors to design 5 experiment programmes which are used to test the improved collaborative filtering algorithm. And then, we test the E-commerce system. The experiment results show that the improved collaborative filtering algorithm can effectively solve the problems of traditional collaborative filtering algorithm, and the accuracy of recommendation results have been significantly improved...