| Recommended system as an important E-Commerce site marketing tool,not only give user valuable advice in the face of mass goods information,but also increased the site sales. Many famous E-Commerce sites have developed recommender system for providing personalization service for consumers. E-Commerce recommender systems are used by E-commerce sites to suggest products for their customers and provide consumers with information that can help them decide which products to purchase.Collaborative filtering recommendation systems have been confronted with many challenges along with the extending of E-Commerce system size. Aimed at the main challenges of E-Commerce recommender systems, this thesis explored and researched recommender systems and some recommendation technologies. At the same time, it analyzed existing problems that collaborative filtering recommendation approach suffered and expatiated on existing resolvents.This paper proposed the model of user's credit,and improved the PIP algorithm.Then it presented a new collaborative filtering recommendation method-CPIP collaborative filtering.The method of collaborative filtering recommendation based on user credit is showed as follows.First of all,this paper carefully analyzed tradtional similarity,then proposed if user rating number more than 30,the similarity of some projects into customers interested in certain similarity and the project not interested, thus promote the formation of high similarity of the neighbor; In the recommended period, designed user credit vector model, make it participate in the recommended, can reduce the influence of malicious rating. The paper makes a detailed algorithm of theoretical analysis, and verifies the rationality and validity of the algorithm is recommended. The proposed algorithm is very good solution to the traditional collaborative filtering method is recommended in the application of existing problems, thus effectively improved the E-Commerce personalized recommendation systems. |
PDF Full Text Request