Design is a hard problem: ill defined and open-ended. Schoen [Schön 83] characterized the process of designing an artifact as a successive refinement of reflection and redesign. Critiquing - the communication of reasoned opinion about an artifact - plays a central role in the design process. A computational critiquing mechanism provides an effective form of human-computer interaction to support these important aspects of design [Fischer 91]. Systems which realize such a computational critiquing mechanism are called Critiquing Systems. These systems provide context sensitive advice and rationale for an artifact designed by a user. This is realized by delivering so-called critiques, which contain relevant information for the user to the task at hand and are some kind of rule of thumb. But design experts are not programmers and programmers are not designers. So we need a module which supports design experts in stating their knowledge in form of critiques. The basis for this module is a a visual critiquing language (here called visual CiLa), completed by a knowledge construction supporting component. Furthermore a single design expert normally does not have all existing design knowledge. So the necessary information for building a complete design system is distributed among different stakeholders. Therefore we additionally need concepts and algorithms to combine and structure the critiquing knowledge of different design experts to construct a trustful, consistent and wise codesigner. This aspect is done by a module constructing the knowledge base and a module for constructing the virtual codesigner. These two aspects - design knowledge construction and presentation - are realized in a tool called Coffein. This article deals with the way Coffein works and how it influences the design process.

visual language, knowledge construction, multi-expert system, scientific visualization, knowledge based sysem, life-long-learning, information systems