It is mentioned that CBR application to design in implementations has resolved theoretical issues. What issues? There may have been discussion of some issues previous to this section but they are not apparent.

      What is a partonomic hierarchy?

      The point about using multimedia representations of information was interesting. It was unclear in the beginning if this meant like audio, visual, etc. or more along the lines of CAD. The examples of systems given later on clarified it to be the prior.

      It wasn't until the Design-case adaptation section that the authors started to present things in a more sensible manner. Most everything before this seemed to ramble on without point.

      The example of Kritik's adaptation methods including component replacement, component replication, and substance replacement meant little in the context of this survey. If these are related or similar to Prompt's Modification Operators, it makes more sense.

      Gencad seems to have some interesting implications. I especially like the idea of not explicitly encoding all the knowledge needed in one specific form dictated by the system but rather gleaning the knowledge through genetic algorithms and possibly neural nets. Most of the system in this field lack from a comprehensive knowledge base needed to solve non-trivial problems. Most of this lies in the knowledge acquisition methods used. It stands to reason that much can be gained from simplifying this.

      Overall, it seems this survey was poorly written. I would have been much more interested in reading more about Gencad. It seems like an interesting system and the author(s) seem far more coherent when discussing it.