扩展产生式规则知识表示方法

Knowledge Representation of Extended Production Rule

通过对现有知识表示方法进行分析,指出了产生式规则和概念图表示方法在自然语言理解中存在的问题,提出了扩展产生式规则知识表示方法,并对该规则的推理机制和具体实现进行了讨论.扩展产生式规则表示方法是将产生式规则的前提、结论和处理分别用概念图表示,是一种混合知识表示方法.它既保持了产生式规则的模块性,又揭示了自然语言中的深层次关系,与自然语言形成自然映射.通过分析表明,扩展产生式规则的总体性能要优于传统的产生式规则,并且特别适合于自然语言理解中的知识表示.该规则已应用于网络故障诊断专家系统的自然语言接口设计中.

A knowledge representation method that can be applied to the natural language understanding and knowledge reasoning conveniently was proposed. By analyzing current methods of the knowledge representation, the shortcomings of the production rule and conceptual graphs were discovered. In order to represent the knowledge in the natural language, the method of extended production rule and its reasoning mechanisms were presented. The extended production rule is a new production rule that the premise, the conclusion and the process of the rule are conceptual graphs. This method is a mixed representation that maintains not only the modularity of the production rule, but also the natural mapping between the natural language and the conceptual graphs. It can also reveal the deep-level relation exactly in the natural language. By analyzing the performance of the extended production rule, it is found that this rule is much better than the production rule, especially fits for the knowledge representation in the natural language understanding. It has been applied to design the natural language interface of the expert system in the fault diagnosis of the network.