Expert systems
They support repeatable decisions when knowledge can be stated explicitly. They can organize diagnoses, check adherence to procedures and suggest investigation steps. They are especially useful when a traceable rule path matters for review.
Text reviewed on
← Back to the 7 pillars
When it emerged
DENDRAL, developed at Stanford from the mid-1960s, is a milestone in expert systems: chemistry knowledge helped form hypotheses about molecular structures. In the 1970s, the approach spread to other domains. The goal was to encode specialist knowledge for bounded problems.
What it is
Represent domain knowledge and use rules or inference mechanisms to support conclusions. They may be small or large and can also combine with methods learned from data.
How it works
A knowledge base holds domain facts and rules; an inference engine combines known facts with conditions such as 'if A and B, then consider C.' A system can show which rule led to a recommendation, but people must build, validate and update its knowledge base. Hybrid systems can combine rules with learned models.
What it is for
They support repeatable decisions when knowledge can be stated explicitly. They can organize diagnoses, check adherence to procedures and suggest investigation steps. They are especially useful when a traceable rule path matters for review.
Where it is used
Equipment fault triage, technical analysis support and product configuration are examples. During maintenance, observed symptoms can activate documented rules that suggest checks without replacing the responsible technician.
Limits and safeguards
Explicit rules do not replace expert review or guarantee correct results outside the intended domain. A rule may be sound while its conclusion is wrong because an input fact was false or the case falls outside its scope. Knowledge bases age; experts must review exceptions, rule conflicts and the consequences of recommendations.