The background and the future of text filtering are described in this paper,and a conceptbased Chinese text filtering model is presented. The main idea of the model is shown as follows: Original keywords are given by users,then the concept expansion is automatically performed with the keywords to construct the user profiles. It is noted that user profiles consist of the subprofiles,and the ratio of sub-profiles matching and the similarity of sub-profiles are defined. As a result,they can weakcn the Boolean constrains to ensure that the text can be matched while some sub-profiles don t match with it,and they can restrict the Vector Space Model to match more subprofiles. In addition,the mechanism of passage matching is applied to improve the efficiency of filtering model.