Diagnostic Reasoning Based on Structure and Behavior,
MASSACHUSETTS INST OF TECH CAMBRIDGE ARTIFICIAL INTELLIGENCE LAB
Pagination or Media Count:
The overall goal of this research is to develop a theory of reasoning that exploits knowledge of structure and behavior. We proceed by building programs that use such knowledge to reason from first principles in solving problems. The initial focus is troubleshooting digital electronic hardware, where we have implemented a system based on a number of new ideas and tools. Troubleshooting digital electronics is a good domain for several reasons. First, troubleshooting seems to be one good test of part of what it means to understand a device. We view the task as a process of reasoning from behavior to structure, or more precisely, from misbehavior to structural defect given symptoms of misbehavior, we are to determine the structural abberation responsible for the symptoms. Second, the task is interesting and difficult because the devices are complex and because there is no established theory of diagnosis for them. Third, the domain is appropriate because the required knowledge is readily available from schematics and manuals. Finally, the application itself is relevant and tractable. Work with a similar intent has been done in other domains, including medicine, computer-aided instruction, and electronic troubleshooting, with the devices ranging from the gastro-intestinal tract, to transistors and digital logic components. Keywords Digital electronic circuits.
- Electrical and Electronic Equipment