扩展产生式规则的网络故障诊断专家系统

根据自然语言理解和网络故障诊断专家系统的特性,提出了一种扩展产生式规则的知识表示方法.扩展产生式规则将规则的前提、结论和建议分别用概念图表示,是一种混合知识表示方法.在扩展产生式规则知识表示方法的基础上,设计出了网络故障诊断专家系统的模型.该系统具有自然语言接口,可将网络中有关故障问题的汉语语句转换成概念图;经过专家系统推理后,可将推理结果和推理过程转换成汉语输出.为了增强专家系统知识获取的灵活性,在系统中还增加了Apriori关联规则挖掘算法,通过对陷阱协议数据单元挖掘,实现在线知识获取.利用该系统模型已开发出网络故障诊断专家系统原型.

Network Fault Diagnosis Expert System Based on Extended Production Rules

According to the characteristics of natural language understanding and expert system of network fault diagnosis,the extended production rule was presented.This rule is a hybrid knowledge representation method that the premise, conclusion and advice of the rules are conceptual graphs. Based on this knowledge representation, a model of expert system with an interface of natural language understanding for network fault diagnosis was designed. It can translate Chinese sentences about faults into conceptual graphs, infer some results from extended production rules, and output Chinese sentences about results and the process of inference. In order to acquire the knowledge flexibility, Apriori algorithm mining association rules from traps protocal data unit of network was added to acquire knowledge online. The prototype of the expert system for faults diagnosis of networks had been implemented in Visual Prolog 5.0 by using this expert system model.