Schema matching plays an important role in various fields of enterprise system modeling and integration, such as in databases, business intelligence, knowledge management, interoperability, and others. The matching problem relates to finding the semantic correspondences between two or more schemas. The focus of the most of the research done in schema and ontology matching is pairwise matching, where 2 schemas are compared at the time. While few semi-automatic approaches have been recently proposed in pairwise matching to involve user, current multi-schema approaches mainly rely on the use of statistical information in order to avoid user interaction, which is largely limited to parameter tuning. In this study, we propose a user-guided iterative approach for large-scale multi-schema integration. Given n schemas, the goal is to match schema elements iteratively and demonstrate that the learning approach results in improved accuracy during iterations. The research is conducted in SAP Research Karlsruhe, followed by an evaluation using large e-business schemas. The evaluation results demonstrated an improvement in accuracy of matching proposals based on user’s involvement, as well as an easier accomplishment of a unified data model.