语境形式化中的演绎与归纳

Deduction and Induction in the Formalization of Context

在人工智能的语境形式化理论中,"局部原则"可以表达为"当前推理活动发生在当前语境当中",它是一条基本原则。在此原则下,语境形式化的过程理应包含两个步骤:(1)确认当前语境的生成要素;(2)建构包含全部生成要素且在演绎规则下封闭的域。显然,第一个步骤是归纳的,第二个则是演绎的。由于缺乏上述区分,造成了已有的理论适用范围受到限制。为应对这种局面而提出的"局部语义"通过将第二个步骤从形式化过程中分离出来,拓展了语境形式化的应用领域。

In formal theories of context in AI, the principle of locality, which can be summarized thus： ＂current ratiocination is in current context＂, is a fundamental principle. Under this principle, the process of formalizing context must involve two steps ： （ 1 ） to identify the generating elements of current context, （2） to build the least do- main which is closed under local deductive rules and involving all generating elements. Obviously, the first step is inductive and the second is deductive. Existing formalizations don＇ t distinguish these two steps and thus their ranges of application are limited. Faced with this, local semantics is proposed, and used to separate the second step from the whole process.