ISHS Acta Horticulturae 313: III International Symposium on Computer Modelling in Fruit Research and Orchard Management
BUILDING EXPERT SYSTEMS FOR USE BY COMMERCIAL FARMERS |
| Authors: | R.M. Crassweller, J.W. Travis, E.G. Rajotte, J. McClure, P.H. Heinemann |
| Abstract:
Expert Systems (ES) are computer programs that have been developed to emulate the logic and problem solving characteristics of a human expert. They are most successful in agricultural applications when addressing specific problems frequently encountered by extension specialists requiring experience, judgement and interaction to arrive at a solution. The advantage is their constant availability, repeatability, dependability and ability to explain the logic used to arrive at the solution. In developing our expert systems we have learned that their are definite common steps that need to be followed in the production of ES. These include the definition of the problem, identification of a suitable expert, extraction of the knowledge necessary to solve the problem and logical arrangement and representation of the expert's knowledge. While traditional methods of ES development require the use of a knowledge engineer (KE) to extract the information from the experts; in the development of our systems the experts themselves served as the KE. The advantage was that it increased the expert's involvement in the development; reduced costs, and increased the ease of use of the systems. We have found that it is best to break the problem down into distinct sub-modules that arrive at solutions that then are fed into other modules. Our systems use dependency networks to organize the knowledge base and employ a frame-based expert system tool, PennShell©. These advantages, disadvantages and common problems encountered in developing the Penn State Apple Orchard Consultant and other systems at Penn State University will is presented in greater detail. | |
Download Adobe Acrobat Reader (free software to read PDF files) | |
