Study Of E-Commerce Recommender System Based On Customers' Interest And Collaboration

As an important marketing tool for the E-Commerce web sites, the recommender system provides customers with valuable advices from millions of item information, so as to help promote the marketing performance of the web sites. With the progress of the Internet and the computer technology, the E-Commerce recommender system receives deeper research and developmentwhich make it produce more intelligent and more personal services. Meanwhile, the scale of E-Commerce is developing so fast that the recommender system is facing a serial of challenges. This paper discusses a few sticking points of the recommender system such as the arithmetic design and its structure.First, the basic theories of the recommender system are introduced, including its background of arising, conception, effect, inputting module, outputting module, some widely used recommending technologies, as well as the hotspots in this field. Second, this paper presents kinds of recomending technologies respectively including information filtering technology, data mining technology, Horting Graph technology etc. and emphasizes the information filtering technology which includes content filtering and collabration filtering. We analyse the signification and the arithmetic of these two information filtering technologies, point out their merit and defect respectively, and then discuss the notion of combining them in the field of recommending the culture items such as movies, music, etc. as this paper `s basic springboard. Third, according to the sparsity of the data and real time problem, this paper promotes the traditional collaborative filtering technology and brings forward a new recommendation mechanism based on the customers ` interests and their cooperations. We describe the principle and procedure of this mechanism taking the example of movie recommending. We use the data provided on MovieLens to carry out the recommending experiment and analyse the Mean Absolute Error. The experiments show that new methods have better precision than the traditional collabrative filtering.Fourth, to make the mechanism more perfect, the value of customers `...