基于有效句型的英汉双语短语对齐

English-Chinese Bilingual Phrase Alignment Based on Effective Sentential-Form

双语短语对齐的输入是源语言 (英语 )、目标语言 (汉语 )候选句法分析树集 由于翻译异常现象的大量存在 ,源语言句法树和目标语言句法树间往往不存在简单的对应关系 如何在翻译异常的情况下 ,从源语言、目标语言候选句法树集中排除歧义 ,选出正确的句法树 ,并揭示其短语层次上的对应关系 ,是双语短语对齐的主要困难 提出的基于“有效句型”概念和“翻译中相对不变准则”的短语对齐模型基本解决了上述问题 双语短语对齐本身是一个复杂度较高的处理过程 ,提出的有关定理和算法使系统在保证对齐正确率的前提下 。

The input of a bilingual phrase alignment system is the sets of source language and target language syntax tree Because of translation divergence, there is usually no simple corresponding relation between source language and target language syntax trees How to pick up correct syntax tree pair from candidate trees and reveal the phrase alignment relation is the main problem The concept and method of 'relative fixed principle of translation' and 'common sentential form' are suitable for this problem The presented theory and algorithm improved the time efficiency