Reasoning visualization in forward-chaining rule-based expert systems
William J. Selig · 1990
A current impediment to understanding the reasoning processes of expert systems is that the information reflecting those processes is either: (a) not available; (b) provided in too much detail; (c) buried within irrelevant information; or (d) all of the above. Due to human cognitive limitations, the task of constructing a mental model of that reasoning from such information is Herculean. Abstract Time Slice (ATS) charts and VISualization Of Reasoning (VISOR) address this impediment by providing pictorial methods for visualizing the reasoning processes of a forward-chaining rule-based expert system based on the information available from a model of those reasoning processes. This dissertation presents: (1) a continuous hierarchical model of the reasoning processes of an expert system application based on a combination of a reasoning model of the host inferencing language and a reasoning model of the abstract application, providing an identification of the information required to be accessible for reasoning visualization; (2) a determination of the capabilities necessary to gain access to this information in a declarative language; (3) ATS charts, which are a static graphical visualization method for the host inferencing language reasoning model; and (4) VISOR, an application level reasoning visualization facility based on the techniques of algorithm animation extended to the more complex domain of reasoning visualization. Prototype visualizations using ATS charts and VISOR are presented to demonstrate the efficacy of the methodology.