This paper describes the use of Web data mining, and analyze the data on the web site (including the server log, commercial database, user database, the shopping cart, user mode) that access to relevant knowledge for goods, commodities such as preference relations. Secondly, the static model of the data mining methods, it is a manifestation of the site management personnel marketing thought. Based on these models, the paper proposed strategy for the site registered users, and produces the corresponding calculating formulas of a good recommendation and the corresponding recommendation algorithm for the current user, thus to get a user recommendation.