Article ID: | iaor1991779 |
Country: | Netherlands |
Volume: | 45 |
Issue: | 2/3 |
Start Page Number: | 275 |
End Page Number: | 292 |
Publication Date: | Apr 1990 |
Journal: | European Journal of Operational Research |
Authors: | Montazemi A.R., Chan L. |
Keywords: | artificial intelligence: expert systems |
An important criterion affecting the success of an expert system is the quality of the underlying heuristics elicited from an expert. It is believed possible though, to use factors such as a formal education, experience, and cognitive functioning to define the level of competence of experts when they perform in a specific domain. Thus, five hypotheses were postulated and tested by means of an empirical study of 17 security analysts in an attempt to study the relationship existing between these factors. It was found that the number of deviation-amplifying feedback cycles in the subjects’ cause maps correlated with the degree of their security price forecasting accuracy. However, no significant correlation between accuracy of forecasting and the subjects’ education, experience, or cognitive complexity was found. In view of lack of an indicator for the choice of an expert with superior performance, application of dialectical inquiry is proposed to assist the knowledge engineer in improving the validity of the underlying knowledge base of the expert systems under consideration.