DEVELOPING TRANSPORTABLE AGRICULTURAL DECISION-SUPPORT SYSTEMS .2. ANEXAMPLE

Information in the area of developing transportable decision support s ystems (DSSs) (agricultural or otherwise) has been scarce. Therefore, developers of DSSs have had some difficulties in constructing DSSs whi ch could be widely used. In the first part of this two-part paper seri es, a conceptual framework was introduced which proposes methods for m aking DSSs transportable. This framework serves as a checklist and inc ludes recommendations about general implementation, user interaction, data management, and model aspects. To illustrate the implementation o f these recommendations, the development of a DSS within a project cal led SYBIL is discussed in detail. Therein, portable and public domain tools have been used to build the DSS with a graphical user interface (GUI) which satisfies the general implementation aspects of the framew ork. Further, the development of a flexible data management system has been essential in the project so that different types of data can be easily handled without changing the models. In order to allow models e mbedded within the DSS to be successfully transported between regions, a novel artificial intelligence (AI) adaptation methodology was imple mented. The main component of this adaptation methodology is a genetic algorithm (GA), an AI search technique. By linking a GA to an agricul tural model, the model becomes more robust because the model is able t o adapt to the region in which it is being used. Overall, by using the framework criteria to select/create these tools/components, the DSS h as been made easy to transport and disseminate.