Intro ] [ 35-Functional Survey ] [ 36-Models ] [ 37-First Principles ]

Up: AI in Design ]

Critique: Context-Dependent Behaviors: A Preliminary Report
Nayak *

      When describing expected behavior I was surprised that it was never referred to as function. While this would not have really changed the concept it might have provided a more direct way to relate this work with other work done on representing and defining (or lack of defining) function of a device.

      When discussing the properties of a good model in the temperature gauge example the idea of the device's function seems to get confused. The authors state that if the gauge's pointer does not move, the device does not measure temperature changes anymore. This is clearly false. The components of the gauge that actually measure temperature may very well be functioning properly. It is the component(s) responsible for displaying the measured value that is incorrect. While this may seem like an unimportant detail or a slight natural language ambiguity, the idea of which components are responsible for what behavior is central to building an accurate model of a device. I was surprised such a mistake was made in this otherwise fine paper. It could be that the expected behavior, relating temperature change directly to pointer position, is what the authors are referring to.

      There seems to be a great potential to exploit this modeling system to provide a form of creative design, or rather, creative reuse of known structures. Given a desired behavior, call it expected behavior if you must, and a library of known structures, the modeling system described in this article could be used to identify those structures in the library that could behave as desired. The further the structure's original intended behavior is from the currently desired behavior, the more creative the solution will appear. Most work, like that done in case-based reasoning systems, focus on finding the closest match to the current problem. This would be at odds with the description of creative given above. It would be interesting to see how a balance could be struck to find a structure that can provide the desired behavior but who's use would still be considered creative.


* P. Pandurang Nayak, Context-Dependent Behaviors: A Preliminary Report. In: Intelligent Computer Aided Design, (Eds) D. C. Brown, M. Waldron & H. Yoshikawa, Elsevier Science Publishers (North-Holland), 1992, pp. 237-251.
Back to Top

 

by: Keith A. Pray
Last Modified: August 13, 2004 8:20 PM
© 2004 - 1975 Keith A. Pray.
All rights reserved.