Abstract: Query rewriting is a basic problem in database research. It is relevant in many aspects of database management, including query optimization, datawarehouse, data integration, semantic cache etc. In order to improve the efficiency of integration system, the frequently submitted XML views are selected to materialize in mediator layer. After end users submit a query, the system tries to use views in mediator layer rather than the access to the data source to answer the query. Such a problem could be reduced to the semi-structured query rewriting. Due to the limited space in the mediator layer, the existing views should answer as more queries as possible. Traditional methods do not take into account the constraints in the semi-structured data, while the constraints play an important role in the transformation of queries hence enhancing the expressive ability of existing views. In this paper, a novel method is explored to make use of semantic constraints, especially path constraints in XML documents to enhance the expressive ability of materialized view in the mediator layer. The soundness and completeness of the algorithm are maintained. The theoretical analysis and prototype preliminary experiment results prove the validation of the method.