一种关于元规则地自然语言生成规则解释技术.pdfVIP

一种基于元规则的自然语言生成规则解释技术* 孙明欣，尹存燕，戴新宇，陈家骏** （南京大学计算机软件新技术国家重点实验室，南京，210093 南京大学计算机科学与技术系，南京，210093） 摘 要：结合一个基于规则的日汉机器翻译系统，给出了自然语言生成规则的一种解释技术。 该解释技术着眼于汉语生成规则语言的表达能力、自然性以及可扩充性，在实现中采用了基 于元规则的方法，使得可以很方便地对规则语言进行扩充；另外，元规则的描述主要采用了 自动机模型。所述的原则和技术也适用于一般的基于规则的自然语言处理系统。 关键词：机器翻译，生成，规则语言，元规则，自动机 中图分类号：TP391 A Meta-Rule-Based Approach to NLG Rules Interpretation Sun Ming-Xin, Yin Cun-Yan, Dai Xin-Yu, Chen Jia-Jun (State Key Laboratory for Novel Software Technology, Department of Computer Science and Technology, Nanjing University, Nanjing, 210093, China) Abstract: Rule-based approaches are still widely used in machine translation even when statistic-based approaches are more and more popular. In a rule-based system, RDLs (rule description language) should be well designed for the convenience of power of expression and extensibility. This paper presents an approach to the implementation of the RDL, which is used for Chinese generation in a rule-based Japanese-Chinese machine translation system. A meta-rule-based interpreting technique is used in the implementation. The meta rules act as the intermedia between Chinese generation rules and the rule interpreter. Thus the updating of the RDL will only result to adding, deleting or changing the meta rules and the interpreter can work well without any modifying. In some meta rules, an automaton model is used to describe some elements of the RDL as complex combinations of atom operations. And other meta rules show that some other elements of the RDL can expressed as linear combinations of atom operations. The principles and techniques discussed in this paper are also applicable to other general rule-based natural language processing systems. Key words: machine translation, generation, RDL (rule des