Our research works are interested in the identification and the representation of the semantic structures of multimedia documents. The semantic structure of a multimedia document is composed of its logical structure as well as the metadata associated to its components. Our goal is to use this semantic structure to store multimedia documents in warehouses. These warehouses must handle and exploit information contained in the relevant documents extracted from disseminated sources. This constitutes a shared repository which must facilitate the cooperative creation and flexible management of the documents. The relational or object database models support with difficulty the semi-structured data and in particular the multimedia documents. Indeed, these documents have complex and heterogeneous structures. Contrary to the databases, the schema preliminary definition for the document databases is not always possible. However, the semi-structured models avoid to impose predefined structures so...