What were the key differences between the DENDRAL and MYCIN expert systems?

The key differences between DENDRAL and MYCIN were in their rule derivation and handling of uncertainty. DENDRAL's rules were based on well-established domain knowledge in mass spectrometry, while MYCIN's rules were derived from expert interviews due to the lack of a general theoretical model. Additionally, MYCIN incorporated a calculus of uncertainty to reflect the uncertain nature of medical knowledge.

DENDRAL and MYCIN were both pioneering expert systems, but they differed significantly in their approach to rule derivation and uncertainty management. DENDRAL was designed to infer molecular structures using domain-specific knowledge from mass spectrometry, where rules were based on well-known patterns and substructures. This made it a knowledge-intensive system with rules derived from established scientific principles. In contrast, MYCIN was developed for diagnosing blood infections and had to rely on rules acquired through extensive interviews with medical experts because there was no comprehensive theoretical model available. Furthermore, MYCIN addressed the inherent uncertainty in medical diagnosis by incorporating certainty factors, a system that quantified the confidence in its diagnostic conclusions.