一个基于GLR算法的英汉机器翻译浅层句法分析器

浅层句法分析是指短语级的自然语言句法分析。在研制MatLink英汉机器翻译系统的过程中,提出了扩充的CFG文法用于描述英语短语句法,并改进了GLR算法,设计实现了用于英汉翻译的英语浅层句法分析器。该分析器采用多出口的分析表结构,引入符号映射函数实现短语边界的自动识别,用孩子兄弟树描述短语的句法结构,并通过短语转换模式实现源语言向目标语言的短语级转换。最后,通过对一个实例句子的分析阐述了该浅层句法分析器的设计思想和工作过程。

A GLR-Based Shallow Syntactic Parser of English-Chinese Machine Translation

Shallow syntactic parsing refers to phrase-level syntactic parsing of natural languages.In developing our MatLink English-Chinese MT system,an extended CFG to describe the syntax of English phrase and a GLR-based algorithm to implement English shallow syntactic parsing have been proposed for English-Chinese machine translation.In this paper,both of them are introduced.There are two multi-exit parsing tables in the shallow syntactic parser.The GLR-based parsing algorithm uses a symbol map function to automatically recognize the boundaries of English phrases and child-brother trees to describe the syntactic structure of phrases.And also phrase transform patterns are introduced in the extended CFG to translate English into Chinese at phrase level.An example is given at the end of this paper to illustrate the designing principles and parsing processes of the parser.