Automating schema matching is challenging. Previous approaches focus on computing all element matches between two schemas and don't take into account the preferences of the user who can be only interested in specific elements of the schema. We propose a new approach based on the user preferences to extract subsets of schemas on which will be applied the matching process. Fuzzy sets can be used to express the user preferences in the selection criteria of a query. Thus, we introduce the notion of fuzzy set defined over a part of the schema, then its extended form that is explicitly defined over the whole schema, according to the generalization rules. This will reduce the research space and therefore contribute to optimize the schema matching process. We also propose to propagate weights to elements of a target schema according to the user preferences on a source schema and mappings found by the matcher between the two schemas. The output scores give an automatic order of the target schema elements based on the interest expressed by the user.