NORM: an expert system for development control in underdeveloped operational contexts

The paper deals with the attempt to build a legal planner expert system, NORM, for the building-application inspection branch of a local city planning agency, within an ill-organized administrative structure. The authors report the first stage of the research, mainly concerned with knowledge acquisition (by means of a method of shared observation, which directly focuses on the expert's real problem-solving behavior), knowledge representation, and implementation of a rough prototype. The first results of the research have shown that heuristics have a strong role in the human expert's cognitive-inferential model, human experts resorting to memory packages, stories, and analogies to streamline their problem-solving, and that many reasoning schemes in the procedure are largely unmanageable by expert systems. Another interesting finding has been the need which emerged from the research for a better understanding of problem-solving routines in the field of planning, ranging from taxonomies to behavioral (especially community-based) patterns in planning.