Application of a network structure in a knowledge-based system for medical diagnostics

Systems supporting diagnostic decision making have been designed more often on the base of production rules than on semantic nets. One reason for that might be that concepts to process information on the latter structures are still rare. In this study the application of a semantic net is described. Via creation of various views, different sets of information are defined, depending on their meaning in the semantic net. Projections, which are in general a combination of mappings and set operations, are used to infer a set of information from another. The developed system was written in Pascal, the semantic net was implemented using a network-like pointer structure. The concept has been used to develop PROJECTOR, an expert system for differential diagnosis based on a semantic net. The structure has turned out to be fast and compact enough to allow large amounts of data to be handled easily even with a personal computer.