Our objective is to make proposals for the design of a SIS SID-quality and meet the needs of different stakeholders of the university. This is where we join (which is poorly modeled by the concept of data marts in the current tools of the market), namely the modeling of data resources. Often the documents are deposited on the information system of an organization without classification, without indexing, with all the information on their content, their purpose, their technical requirements and practices. The method of describing the properties of a document is a binding step involves an author and a culture of destruction of documents. Few users perform document properties they file on a system design and information. Then it is naturally more difficult to retrieve these information gaps which usually take the form of voids, it is still necessary that the input fields are provided adequate and appropriately organized, arranged and explained. Indeed, it often happens - for example on an intranet of an organization - the drop zones are not conducive to give relevant information on the properties of materials downloaded. In the best case, the documents are managed by their own systems, accessible through their own search engine or by federated search engines. Why we try to answer the question: how to reproduce a set of metadata specific to multidimensional databases specific to the decision-oriented universities.