Here we see again the idea that more knowledge stored in a design case, the better. This approach adds knowledge in the form of a qualitative model. Also present in this approach is a high degree of cross-indexing between the different kinds of knowledge stored in a case and the qualitative models and modification-generation plans in the system.

      The main mechanism for modifying or adapting design cases is the use of families of modification-generation plans. It is unclear from where these plans originate though. One can only assume they are generated manually by a domain expert.

      I was disappointed that the method by which cases are retrieved was not discussed in any detail. This seems pivotal to the operation of the system, especially from the model-based case adaptation point of view. A case should be retrieved that is the best fit for the new design problem at hand. This measure of fit must take into account the differences between the function(s) provided by the retrieved design and the function(s) required by the new design problem. The identification of these differences leads then to selecting/generating a plan to modify the retrieved case to fit the new problem. Yet there is no discussion past stating a case is retrieved. This is an integral part, as it is in all means-ends analysis techniques.

      This article does illustrate the power and utility of using qualitative knowledge and functional reasoning within a case-based reasoning design system. By not relying solely on the structure of a design but including its purpose, how the stored designs provide function, the effort needed to solve a new design problem can be better focused to achieve results more efficiently.