逐步求精法获取上下文无关文法  

A STEPWISE REFINING METHOD FOR INFERRING CONTEXT\|FREE GRAMMARS

作　　者：张瑞岭[1] 董韫美[1] 

机构地区：[1]中国科学院软件研究所计算机科学开放研究实验室,北京100080

出　　处：《计算机研究与发展》2000年第1期30-36,共7页Journal of Computer Research and Development

基　　金：国家自然科学基金!(项目编号69673018)

摘　　要：文法推断研究如何从语言的有限实例，通过归纳推断获取语言的文法定义． 文中提出一个基于逐步求精的上下文无关文法推断方法，以尝试将文法推断用于替代或帮助传统手工的文法构造工作． 文中的推断方法以Ａｎｇｌｕｉｎ 的交互式学习模型为框架，以逐步求精和复用为主要策略。Grammatical inference(GI) is a problem of learning languages from examples where the target domain is a formal language and the representation class is a family of grammars. A constructive method for inferring context\|free grammars from positive examples is presented in this paper. This work is actuated by the intention to apply GI to replace or supplement the conventional manual grammar construction. The inferring method has the following features: (1) it is interactive; (2) it is incremental; (3) it employs such strategies as stepwise refining and reuse to enhance its learning capability. and (4) the inferred grammars have natural structures.

关 键 词：归纳学习 上下文无关文法 逐步求精 形式语言 

分 类 号：TP301.2[自动化与计算机技术—计算机系统结构]