In this paper we present the GENCAD expert system which helps solve problems involving the structural design of tall buildings, pecifically focusing on the ways that the buildings resist lateral movements. One of the features of this type of design problem is that there doesn’t seem to be a lot of knowledge that can be made explicit about solving problems in the domain in order to program an expert system with it. Whether this lack of explicit knowledge is due to the complexity of the designs involved; to the implicit, intuitive, nature of the knowledge used by human experts; or to other factors, it still hinders the construction of a robust knowledgebased expert system. Our methodology has therefore been to make GENCAD a knowledge-lean expert system. This has been achieved by performing a careful task analysis of the problem-solving task and domain, and by implementing the required subtasks as much as possible using artificial intelligence methods that require little or no explicitly-coded domainand task-specific knowledge. We present here our analysis and our implementation of GENCAD.
[1]
Nigel Cross,et al.
Elicitation of Rules for Graphic Design Evaluation
,
1996
.
[2]
Andrés Gómez de Silva Garza,et al.
The Adaptation of Structural System Designs Using Genetic Algorithms
,
1996
.
[3]
I. D. Bennetts,et al.
Structural Systems for Tall Buildings
,
1995
.
[4]
Andrés Gómez de Silva Garza,et al.
Design by interactive exploration using memory-based techniques
,
1996,
Knowl. Based Syst..
[5]
David E. Goldberg,et al.
Genetic Algorithms in Search Optimization and Machine Learning
,
1988
.