Uncertainty in Expert Systems with Multiple Sources of Information Application to Agricultural Production

该建议将统计建模和模拟的强大能力与基于规则的专家系统的开发相结合，这将有助于解决农业生产问题。在专家系统模型的背景下，期望在开发更好的方法来处理和表达不确定性方面取得总体进展。在开发保护大豆作物免受虫害的改良工具（这是农业生产研究中的一个重要问题）以及在将决策分析技术的使用扩大到农业部门方面预期会取得具体进展。该项目将采用贝叶斯概率推理框架，并通过影响图表示随机推理网络，这将有助于构建问题并探索关键变量之间的依赖关系。该项目的成功应该有助于证明使用基于计算机的专家系统作为决策辅助工具来解决涉及风险和不确定性的生产决策的可行性。

This proposal combines strong capabilities in statistical modeling and simulation with the development of rule-based expert systems that will assist in the solution of agricultural production problems. General progress is expected in developing better ways to handle and express uncertainty in the context of expert systems models. Specific progress is expected in developing improved tools for protecting soybean crops from pests, an important issue in agricultural production research, and in extending the use of decision analytic techniques to the agricultural sector. The project will employ the framework of Bayesian probabilistic inference and the representation of stochastic inferential networks through influence diagrams, which will help to structure the problem and explore dependencies between key variables. The success of the project should help to demonstrate the feasibility of using computer-based expert systems as decision aids to address production decisions involving risk and uncertainty.