Modeling human intuition in an expert system for planning aerial application operations

Abstract

Acquisition of expert human knowledge is a critical aspect of the formulation of computerized expert systems. Some aspects of the process of human reasoning do not land themselves in any forthright manner to current techniques of computer programming. Simulating, in computercoded form, the intuitive nature of human-deductive reasoning is perhaps the most challenging difficulty in developing a successful expert system. The ability of the human expert to define his (or her) decision making in terms of the system's knowledge representation scheme is quite limited in most cases. Even the most cooperative expert will occasionally find that some of his reasoning cannot be adequately described by the relatively rigid framework of the system being developed. One method of overcoming this problem is through the incorporation of heuristic models. These models essentially serve as ‘patches’ in an otherwise continuous representation of the problem space and the human expert decision-making process. Examples are shown, taken from an actual aerial application operations case where heuristic models were used successfully to capture knowledge claimed by the human expert to be of an ‘intuitive’ nature. Integration of these models into an otherwise rigid framework has considerably simplified the knowledge acquisition process without compromising the overall performance of the system.