“Rule-Based Expert Systems: The MYCIN Experiments of the Stanford Heuristic Programming Project,” edited by Bruce G. Buchanan and Edward H. Shortliffe, was published by Addison-Wesley in 1984. It is now out of print, and Shortliffe has made all chapters freely available on his own website, which is the copy cited here. The book is a critical, chapter-by-chapter analysis of nearly a decade of research centered on MYCIN, the medical expert system Shortliffe began as a Stanford doctoral project to advise physicians on antibiotic therapy for bacterial infections.
The volume is the most complete account of how a rule-based expert system actually works. It covers the structure of MYCIN’s rule base and how knowledge was extracted from human experts, the certainty factor model the team devised for reasoning under medical uncertainty, and the system’s ability to explain its reasoning by tracing back through the rules it fired. It also documents EMYCIN, the effort to strip MYCIN’s medical content out and leave a reusable shell for building other rule-based systems, an idea that helped launch the commercial expert-systems industry. Later chapters honestly assess MYCIN’s performance evaluations, its tutoring spinoffs, and the practical obstacles, including why MYCIN was never deployed in clinical practice.
Why a business reader should care: this book is the engineering bible of the expert-systems era, and its treatment of knowledge acquisition, uncertainty, explanation, and evaluation maps directly onto the same hard questions facing any organization deploying a knowledge-based or AI decision system today.