In a cycle of research, in which each scientific document, itself inspired by previous papers, will be read and commented by researchers who will quote it, a document will be an alternately finished, an updated production and a potential source of appropriation and quotation. After an introduction to information retrieval, we will examine the typology of scientific documents, what the storage and broadcasting of information model can be, as well as their standards and related protocols. We describe several methods of document retrieval and various tools of interrogation of knowledge bases. We postulate that nowadays, the document retrieval can be technically automated, since the first stage of establishment of the research's scope up to the writing of the bibliography. We introduce dedicated tools and standards can also facilitate the stages of selection and documentary management. We propose a user focused study in order to establish several users' profiles and associated uses. Then we submit a conceptual process of visual information retrieval. To do so, we analyze methodologies of rating this specific type of paper as well as scientific papers' criterions and standards. We synthesize all the bibliographical procedures of automation to model a search tool allying respect for the standards, the flexibility of use and consideration of the of the users' cognitive needs. This interface will be used as support to browse conceptual domain as an access to scientific corpuses. It will also expose of document's metadata with LIS good practices and semantic web's rules. After evaluation, this tool is generalized as a model for scientific information retrieval design.
