Research On The Recommender System Based On C2C E-Commercial Enviroment

Currently, the development of E-commerce is under rapidly growth in our country. With the number of users of internet and consumer who participate in E-commerce is growing larger and larger. The scale of E-commerce is under rapidly growth. There are many various kinds of commodities open up before the consumers' eyes in E-commercial environment, but they always feel confused and could not find the really thing that they want because there is no recommender services which is offered by salesmen in shopping mall. In this condition, the recommender system in E-commercial system is very important for consumers to find what they really want. So there many experts and scholars do research on recommender systems. There are many researches based on B2B and B2C E-commercial environment but it is rarely to find some researches based on C2C E-commercial environment.In this dissertation, based on previous research, integrate with the characteristic of consumer behavior in E-commercial environment, do the research on recommender systems based on E-commercial environment through empirical and experimental analysis. Firstly, chapter one and two organizes and summarizes the definition of recommender system and the behavior of network consumers, the classification of recommender method, the characteristic of network consumers. Secondly, on the basis of literature review and theoretical analysis, integrate with the process of network consumers' decision. There are ten impact factors are selected for further research. Then three impact factors are identified for the further research on recommender system through the empirical analysis. Finally, on the basis of collaborative recommender method, integrate the network consumers'impact factors, the modified method is explored and further more probabilistic model has been led into recommender method. And the probabilistic model recommender method based on C2C E-commercial environment is explored. Then there is a comparison among these three methods has been given by experimental analysis. The results reflect that the recommender quality of modified recommender method is better than the previous one.With the effect of impact factors and probabilistic model's introduction, the recommender method will perform better to find the latent reason why consumers purchase the commodity. And finally it makes the recommender method satisfy consumers' needs much better. It has significant contributions to the development of E-commerce.