An `AI in Design' View of Design

Internal Evaluation

Does the system provide any internal critiquing, evaluation, estimation, or advising methods?
What other kinds of internal feedback exist in the system?

Discussion. These methods provide types of meta-knowledge that support making decisions. They are important in knowledge-based systems, where the design values usually do not result merely from a simple computation. Estimates give an approximate value for design parameters. Evaluations gather information about a parameter value or a design decision and construct an approximation of its quality. Critiques (both positive and negative) comment on a design entity (parameter value, decision, etc.) from a specific point of view [Grecu & Brown 1996]. Advising gives suggestions about possible solutions.

All these forms of meta-knowledge generation become particularly important if the design is carried out by several agents with different perspectives [Werkman 1992]. They also act as feedback during design reasoning. As design is a complex task, feedback information often becomes available remotely from the place where it is needed. In such cases, design systems need special mechanisms to connect the places where the meta-knowledge is generated to the places where it is needed [Marcus et al. 1992] [Stefik 1981b].

Up : Next

/Papers/IJDC/sect5.13.html Wed May 28 14:46:09 EDT 1997