Choosing plants for a garden can depend on climate, soil, and sunlight. An expert system aims to make that kind of domain-specific decision using knowledge gathered from human experts. It separates that knowledge from the procedure that applies it, so decisions need not involve an expert every time. section 2.1
Three components work together:
- A knowledge base stores expertise as if-then rules. section 2.1
- Working memory holds input facts and tracks what has been deduced. section 2.1
- An inference engine repeatedly checks facts against rules. Triggered rules can add or change facts; the loop stops when no further rule can trigger. section 2.1
A medical reasoning example
One illustrated rule assigns a 40% likelihood of a particular organism when several patient and laboratory conditions hold. A broader diagnostic outline prioritizes emergencies, then obvious diagnoses, then reasonable tests for likely explanations. These examples show how expert knowledge can guide decisions, not that a system replaces clinical judgment. section 2.1 section 8.0
Separating rules from the inference procedure lets domain experts contribute knowledge and makes rules easier to add, update, or remove. It also puts the quality and suitability of those rules at the center of the decision. Strict true/false distinctions can be limiting for nuanced questions; fuzzy logic addresses that limitation. section 2.1 section 2.6
