Data management can be achieved by different types of systems: from centralized file management systems to grid and P2P systems passing through distributed systems, parallel systems, and data integration systems. An increase in the demand of data sharing from different sources accessible through networks has led to proposals for virtual data integration approach. The aim of data integration systems, based on the mediator-wrapper architecture, is to provide uniform access to multiple distributed, autonomous and heterogeneous data sources. Heterogeneity may occur at various levels (e.g., different hardware platforms, operating systems, DBMS).