The main idea in this article is that groupware benefits organizational memory because it focuses on communication and coordination, but is inadequate for an efficient knowledge management. It has to be completed by a model for exchange structuring in order to improve dialog quality and to enable a conversation classification that would not be simply chronological. Our aim is to build a groupware (MEMO-Net) enriched with such a model. This model (DIPA), which uses and simplifies the concepts of Problem-Solving methods, comes from a review of existing Design Rationale formalisms that gave rise to the ABRICo formalism. 1. Introduction The result presented in this article belongs to a research project whose aim is to manage knowledge used in design project for capitalization and reusability. In accordance to Zacklad and Grundstein ([24]), knowledge capitalization research can be classified in three categories: social and cooperative approaches, top-down modeling approaches and bottom-up modeling approaches. In the first category, one considers that organizations' critical knowledge comes within a collective competence that is not enough or bad formalized. The development of systematic use of groupware, electronic-mail services, newsgroups, workflow, embodied in Intranets and particularly used in design projects would seem to explain that these tools are indeed considered potential aids for the knowledge capitalization process. If groupware mediatizes designers' interactions, the best way to locate and protect crucial knowledge (in the sense of M. Grundstein, [13]) exchanged through these nets is to study the interactional structure and to suggest information structuring tools and models for highlighting exchanged knowledge and enabling easier future access. Most of the time i n knowledge capitalization projects, there is a lack of quality rather than quantity of information; quality regarding the structure of memorized materials. In "social and cooperative" knowledge capitalization approaches, there seem to be two types of approach. Some work will aim for a posteriori structure information by reconstituting, from the traces of intellectual transactions ([25]), the concept structure previously elaborated collectively. Others will aim for a priori structure transactions to guarantee a better quality of both interactions and write track of these interactions that will enable an easier re-exploitation by the "knowledge managers".