A software system (SAGE) was built for on-farm decision support. The objective was to provide a framework for constructing and deploying knowledge-based decision support in the areas of integrated pest management, fertilization, and field operations. The framework is open by design and includes a generic model of an agro-ecosystem as well as various mechanisms allowing for the continued growth in scope and function of the software. The SAGE system is designed to provide a number of building blocks and predefined decision-support strategies that can be adapted to specific needs and situations. It operates on a personal computer and is based on the use of an objectoriented technology for software construction and operation. A prototype of the system has been built and is being used to build commodity-specific decision-support modules.
In Quebec, the carrot Cercospora blight represents a major foliar disease. In carrot fields, it causes reductions in yields of up to 30%. The evolution of this disease can be predicted by considering the meteorological and biological parameters and by using expert knowledge. Disease management can be enhanced through the use of a computerized decision support system (DDS). The objectives of the project were: 1) to define a conceptual framework for the operation of a carrot protection module, 2) to integrate a model of Cercospora blight evolution within the framework, 3) to integrate and structure the information needed for the consultation of the DSS, and 5) to validate the recommendations of the module. The various components (knowledge base, database, simulation model) constitute an extension to an existing framework used for agricultural production management (SAGE). The latter is built using an object-oriented programming language (Smalltalk) and an object-oriented database management system. A fully operational version of the system was developed and will be tested during the summer of 1995. The developed system combines a Cercospora blight model and a set of rules that convey the expert's knowledge. These rules were formulated based on interviews with the expert. The nature and organization of the rules will be presented as well as a critical evaluation of the methodology and tools used to build the system.