The author presents a higher-level technique for causal reasoning and diagnosis. Modules are the basic components in the resulting representation. Considerations unique to causal reasoning at a high level are discussed, and an algorithm and examples incorporating the resulting characteristics are presented. Using higher-level causal reasoning has been shown to be a useful technique for introducing the flexibility of causal reasoning to complex systems for which module-level diagnosis is desired. A prototype system, FEXPR, has been successfully implemented to demonstrate the higher-level causal diagnosis approach.
展开▼