Combining experiential and theoretical knowledge in the domain of semiconductor manufacturing
From Semantic Portal Wiki
{{#vardefine:category|Publication}}{{#vardefine:templatename|i.publication}}{{#vardefine:package|smwbp_instance_templates}}
| Edit |
Reference: {{#vardefine:pagename|combining experiential and theoretical knowledge in the domain of semiconductor manufacturing }}
- [[]]
bibtex
{{#vardefine:pagename|Combining experiential and theoretical knowledge in the domain of semiconductor manufacturing }}{{#vardefine:key| }}
abstract: Semiconductor Manufacturing is characterized by complexity andcontinual, rapid change. These characteristics reduce theeffectiveness of traditional diagnostic expert systems: the knowledgerepresented cannot adapt to changes in the manufacturing plan becausethe dependence of the knowledge on the plan is not explicitlyrepresented. It is impractical to manually encode all thedependencies in a complex plan. We address this problem in two ways. First, we employmodel-based techniques to encode theoretical knowledge, so thatsymbolic simulation of a new manufacturing plan can automaticallyglean diagnostic information. Our representation is sufficientlydetailed to capture the plan's inherent causal dependencies, yetsufficiently abstract to make symbolic simulation practical. Thistheoretical knowledge can adapt to changes in the manufacturing plan.However, the expressiveness and tractability of our representationalmachinery limit the range of phenomena that we can represent. Second, we describe Generic Rules, which combine theexpressiveness of heuristic rules with the robustness of theoreticalmodels. Generic Rules are general patterns for heuristic rules,associated with model-based restrictions on the situations in whichthe patterns can be instantiated to form rules for new contexts. Inthis way, theoretical knowledge is employed to encode the dependenceof heuristic knowledge on the manufacturing plan.
download:
- paper:
- slides:
| Abstract | Semiconductor Manufacturing is characteriz … Semiconductor Manufacturing is characterized by complexity andcontinual, rapid change. These characteristics reduce theeffectiveness of traditional diagnostic expert systems: the knowledgerepresented cannot adapt to changes in the manufacturing plan becausethe dependence of the knowledge on the plan is not explicitlyrepresented. It is impractical to manually encode all thedependencies in a complex plan. We address this problem in two ways. First, we employmodel-based techniques to encode theoretical knowledge, so thatsymbolic simulation of a new manufacturing plan can automaticallyglean diagnostic information. Our representation is sufficientlydetailed to capture the plan's inherent causal dependencies, yetsufficiently abstract to make symbolic simulation practical. Thistheoretical knowledge can adapt to changes in the manufacturing plan.However, the expressiveness and tractability of our representationalmachinery limit the range of phenomena that we can represent. Second, we describe Generic Rules, which combine theexpressiveness of heuristic rules with the robustness of theoreticalmodels. Generic Rules are general patterns for heuristic rules,associated with model-based restrictions on the situations in whichthe patterns can be instantiated to form rules for new contexts. Inthis way, theoretical knowledge is employed to encode the dependenceof heuristic knowledge on the manufacturing plan. istic knowledge on the manufacturing plan. |
| Author | John L. Mohammed + |
| Bibtype | techreport + |
| Institution | Stanford University + |
| Key | KSL-94-62 + |
| Note | STAN-CS-94-1526 September. + |
| Number | KSL-94-62 + |
| Tag | Computer science + |
| Title | Combining Experiential and Theoretical Knowledge in the Domain of Semiconductor Manufacturing + |
| Tr id | KSL-94-62 + |
| Year | 1994 + |
