expertsystem

An expert system is a computer program that emulates the decision-making ability of a human expert. It is designed to solve complex problems in a narrowly defined domain by applying a store of domain knowledge and a set of reasoning rules. The term expert system is widely used in English; "expertsystem" appears in several languages as a direct translation.

A typical expert system comprises three core components: a knowledge base that contains domain facts and heuristics; an inference engine that applies the rules to the knowledge to derive new information or decisions; and a user interface for interaction. Some systems include a knowledge acquisition component to capture new knowledge from experts.

Knowledge is usually encoded as if-then rules, but can also be organized as frames or semantic networks. The inference engine often uses forward chaining (data-driven) to reach conclusions or backward chaining (goal-driven) to justify a particular hypothesis. The RETE algorithm is a common technique to optimize rule matching.

History: The concept emerged in the 1970s and 1980s with projects such as DENDRAL (chemistry, hypothesis generation) and MYCIN (medical diagnosis). Expert systems were among the most successful AI applications of their era, finding use in medicine, engineering, geology, finance, and troubleshooting.

Advantages include the ability to codify expert knowledge and provide explanations for its conclusions. Limitations include brittleness, the knowledge acquisition bottleneck, difficulty adapting to new domains, and limited learning capability. Modern AI often emphasizes data-driven approaches, but expert systems persist in niche applications and hybrid forms, frequently integrated with machine learning or rule-based shells (such as CLIPS or JESS).