I. Introduction
Schema matching refers to get the mapping relation between elements of the two given schemata. As is shown in Figure 1, we give an example of schema matching. Schema matching is always an important issue in many fields, such as data integration, data warehouse and electronic commerce. A lot of significant researches have been proposed, and some of the ideas have been implemented as schema matching tools. COMA [8], iMap [7] and Clio [6] are classical among the proposed methods, and these methods are respectively based on the schema semantics or schemata instance information. It cannot be denied that these excellent methods can give comparatively accurate answers, but deficiencies inevitably exist. The accuracy of these methods is generally low, namely the high uncertainty that exists in other fields [13], [14]. The prime causes of the uncertainty's occurrence is the fuzziness of data's descriptions between two heterogeneous databases and the ambiguity of schema's descriptions between the source schemas and the target schemas [13] [14].