The Research Of Recommendation System In E-Commerce Based On Web Data Mining

It's hard to preserve customers in Electronic Commerce, and more difficult to gain new customers also. So setting up a good relation with customers, improving customers' loyalty becomes more important. So a recommend-system which can provide recommending-service according customers' favorites, behavior information is necessary. That is network "one to one marketing". The paper establishes a recommendation-system model according efficiency and veracity; and describes the each module in the model exactly, especially on how they correspond with each other. The model uses both web usage minding and web content mining to analyze and set up the pages, products and customers pattern with server–data which is used as accordance of recommendation; then describes the strategy and arithmetic that we use in the model, and import a new index which we call average visitation degree to improve the accuracy; at last gives out the result of test with real data and try to explain it.