This paper presents a domain-oriented user attribute model for getting user personalized and real requirements on the field of the accurate retrieval. Resource Distribution Matrix (RDM) and User�s Personalized Preference Vector (UPPV) are created by analyzing the user feature and resource characteristic on domain � specific condition. We then present a method to improve the efficiency of the personalized accurate retrieval with the user attribute model. Finally, the preliminary experiments with the prototype indicated that there was an increase in the precision of individualized resources retrieved as the model combines with Shanghai Education Resource Library.