摘要
把面向对象技术应用于专家系统知识库和推理机的建造,提出了一种基于规则推理网络模型优化搜索的设计思路,并讨论了系统的软件实现技术。推理模型由数据驱动,采用广度优先遍历模型推理网络,可降低搜索次数。将其应用于吉林省农作物熟区动态区划专家系统的开发中,采用模块化设计,推理过程单独设计一个推理类,比传统的推理机节省了一半以上的搜索次数,取得了很好效果。在实例应用中采用定性与定量相结合的综合多级推理,该方法降低了问题求解的难度,更加贴近人类专家的思维过程。
Using object-oriented technique in building knowledge base and rule reasoning model of expert system,a new method based on rule reasoning network model which has better search path is proposed, and software realize is discuss ed. Reasoning model is activated by data. Reasoning times can be reduced by using wide, bette r and network. Using this means in 'cropper mature area dynamic subarea expert system of Jilin province' and getting very good effect, system is designed in the module. Reasoning process is designed as a independent class, so ,the search times is as half as, even less than the one of classical reasoning machine. In the application we use integrating multilevel reasoning which combined qualitati ve and quantitative reasoning, it degrade the difficulty and more close to the t hinking process of human domain expert.
出处
《吉林大学学报(信息科学版)》
CAS
2003年第2期158-161,共4页
Journal of Jilin University(Information Science Edition)
基金
吉林省科技星火计划项目(2001EA660024)
关键词
专家系统
面向对象
知识表示
规则类
Expert system
Object-oriented
Knowledge presentation
Rule class