Download Research Summary_

From: AAAI Technical Report FS-92-03. Copyright © 1992, AAAI (www.aaai.org). All rights reserved.
Research Summary_
Nameand Address:
Dr. Sattiraju Prabhakar
School of ComputingSciences
University of Technology, Sydney
PO Box 123, Broadway
Sydney, NSW2007, Australia
email: [email protected]
Tel. (O): (61-2) 330-1851
Fax: (61-2) 330-1807
The primary goals of this work are to develop (i) the structure-behavior-function (SBF) modelsfor various kinds
of knowledgeused by the designer during Creative Design, and (ii) the methodologies to composedifferent, but
related, SBF models to arrive at reusable design models. Wehave been developing a methodology called
PerformanceDriven Creativity (PDC)which arrives at creative designs by considering the interactions of existing
design knowledgeof a device and the failure knowledgeof this device in a new operating environment [Prabhakar
and Goel 1992]. The failure and the design knowledgeare represented using different SBFmodelswhich use similar
ontologies. The failure knowledgecan be arrived at using (i) the previous experiences of design failures, (ii)
prototypical behaviors, or (iii) basic concepts in the failure SBFmodel[Prabhakar 1989, Prabhakar, et al 1990]. The
resulting failure knowledgerepresents howa fault develops, while the device performs within a new operating
environment,due to the existing design knowledge.Oncethe failure knowledgeis arrived at, it is used to discover
the new constraints the knowndesign of the device should follow. Newbehaviors are generated for these
constraints. These behaviors are composedwith the behavior of the device. Since the behaviors that realize the new
constraints can be significantly different from the knownbehavior of the device, composition of these behaviors
needs comprehension, abstraction, specification or reorganization of the concepts involved in describing the
behaviors. Someof the methodologies for behavior composition have already been identified. Further work is in
progress to identify and develop other methodologies.
[Goel and Prabhakar 1991] A. Goel and S. Prabhakar. "A Control Architecture for Model-basedRedesign Problem
Solving". Artificial Intelligence in Design Workshop,Proc. Twelfth International Joint Conferenceon Artificial
Intelligence, pp. 121-136,Sydney,Australia, Aug.1991.
[Prabhakar 1989] S. Prabhakar. "Diagnostic Reasoning with Surface and Deep Level Knowledgein Medical
Domain".Ph.D Dissertation, Dept. of Electrical Engineering, Indian Institute of Science, Bangalore, India, 1989.
[Prabhakar and Goel 1992] S. Prabhakar and A. Goel. Using diagnostic experiences in experience-based innovative
design. In Proceedingsof Applications of Artificial Intelligence X: Knowledge-Based
Systems, pp. 420-434, 1992.
[Prabhakar, et al 1990] S. Prabhakar, I.S.N. Murthy and L.M. Patnaik. "Diagnosing Multiple Faults using a
Taxonomic Organization of Deep Level Causal Knowledge". AAAI Spring Symposium Series: AI in Medicine,
Stanford University, 1990.
118