Data integration involves combining data residing in different sources and providing users with a unified view of these data through what is called a "global schema''. The authors address here the problem of the construction of this global schema, with a minimum human effort, in the semantic Web context where data sources are annotated with ontologies. The authors aim to facilitate the task of building a common vocabulary (ontology) that will serve as a shared conceptual level for several heterogeneous data sources needing to share their data in a specific application domain. The authors propose a solution based on the use of a domain reference ontology (or "background knowledge'') as a mediation support and some reasoning techniques in order to minimize human intervention. The work presented here is implemented as a prototype for Semi Automatic Global Ontology Building (SAGOB).