摘要
在以规则为基础表示知识的专家系统中,推理机制通常都是采用“向前推理”,“向后推理”或者“混合推理”。这些推理方式随着规则数目的增多,推理过程的匹配运算呈“组合膨胀”之势,以致极大地影响推理速率和专家系统的实用性。本文提出的一种称之为“决定式推理方式”的推理方法,圆满地解决了传统推理方式上存在的组合膨胀问题。
In the conventional rule-based system,several different reasoning methods
are adopted,which take much time.This is a serious problem when the rule-base
becomes very large.
This paper gives a new type of reasoning method in expert system,which
is called deterministic reasoning.In this system,all rules in knowledge base are
pre-compiled into a knowledge net and a mixed data-driven/object-driven control
structure is used.The system can raise efficiency greatly.
基金
国家自然科学基金
关键词
专家系统
决定式
推理
人工智能
artificial intelligence
combination
expansions/deterministic reasoning
knowledge base
