嵌入元级结构的基于事例推理研究

THE STUDY OF CASE-BASED REASONING WITH META-ARCHTTECTURE

元级推理是智能软件系统中的一种重要机制。本文首先讨论了元级推理的作用、原理以及实现策略,然后重点介绍了我们新近完成的一个面向事例诊断的推理器MGCBD。本文各节介绍了MGCBD的知识表示、元级推理框架以及事例检索、匹配和适配过程的设计方法和技术特点,并用一个具体实例来说明MGCBD的推理功能。在本文的最后部分,我们将MGCBD与国际上的同类系统进行了对比和分析。

Meta-reasoning has been regarded as an important mechanism in current knowledge-based systems, this paper firstly discusses the principles of meta-reasoning, then describes the roles and practical methods of embedding meta-reasoning into the ground-level reasoning. The paper focuses on our newly developed Metaobject-Guided Case-Based Diagnostician (MGCBD). And we give a detailed description to its knowledge representation, meta inference model, and specific implementation strategies relating to the case retrieval, matching and adaptation sub-process. Finally, this paper compares our system with other similar systems in the world.