The knowledge engineering and software development tasks involved in constructing software for an expert system are very different from the task of designing an imperative program. Some of the problems in the construction of a rule-based expert system include deducing the heuristics of the expert, converting these heuristics into a working taxonomy and rule base, and ordering the rule base so that the system performs efficiently and correctly. A pictorial method of knowledge representation that we call a "k-tree" greatly facilitates these tasks.
/lp/association-for-computing-machinery/a-pictorial-aid-for-programming-expert-systems-4t10n7Pigr