Definition
An expert system is a knowledge-based AI architecture consisting of a rule base (if-then rules encoding expert knowledge) and an inference engine that applies the rules to specific cases. Prominent examples include MYCIN (medical diagnosis, 1972), DENDRAL (chemical structure, 1965), and XCON (computer configuration, 1978). The architecture was the dominant commercial AI approach in the 1980s.
Why it matters
How it works
Build a rule base by interviewing domain experts and encoding their knowledge as IF-THEN rules. Encode each case as a set of facts. The inference engine applies forward chaining (from facts to conclusions) or backward chaining (from goals to required facts), firing rules whose conditions are satisfied. The system can usually explain its reasoning by listing the rules used. Limitations: the rule base is brittle outside the elicited cases, hard to keep current, and labor-intensive to extend.