The information explosion combined with the development of communication technologies, have made difficult content access and exploitation of the scientific and technical mass of documentation. This constitutes a discouragement reason for the users. For this reason, the control of information access process remains a problematic and a major stake for the scientists and the information professionals. This leads us to resort to the information organization and sense management tools, rising from a major importance. It is in this context that a particular interest is accorded to the use of graphs theory in information science field and particularly for the information cartography. The principle of such cartography is to enable the users to take into account the existing links between the different notions processed in a selected textual subject corpus. Thus, the corpus takes the form of a card called semantic card, which represents the processed subject, so enabling the user to achieve a best comprehension of its structure. In order to achieve this objective, we suggest an approach for the development of a three dimensional cartography: global, by communities and punctual (specific). The suggested method lies on the structural properties of the terms association network model, as well as the notion of representative generic set. This set could efficiently be found on a polynomial time.

information cartography, textual corpus, representative generic set, network of association of terms.