A systems analysis approach to brown planthopper control on rice in Zhejiang Province, China. III. An expert system for making recommendations.
Abstract
An expert system was constructed to help diagnose problems caused by the delphacid rice pest Nilaparvata lugens in Zhejiang Province, China, and to determine insecticide recommendations. The information and knowledge used in the expert system was obtained from 3 sources: the expert opinion of one of the authors (J.A. Cheng); the results of simulation modelling (J.A. Cheng & J. Holt (1990) and J.A. Cheng, G.A. Norton and J. Holt (1990)); and information available at a regional level (e.g. weather forecasts and pest forecasts) and field level (e.g. rice variety and transplanting time). The logical structure and knowledge base of the expert system was encoded in a software package (or shell) for use on IBM-compatible machines. Preliminary tests of the expert system indicated that it provided advice on insecticide treatment that was at least as good, if not better, than that provided by the rules used by 2 groups of extension officers. It is suggested that the value of this expert system lies in two areas. Firstly, as a user-friendly manual and training tool, it can improve the recommendations that agricultural advisers make, and provide additional information on N. lugens and its control. Secondly, as a means of increasing communications between research and extension staff, the expert system can help in the identification of research, surveillance and monitoring needs.