基于UAMRT的机器翻译方法

本文提出了一种新的机器翻译方法，即基于UAMRT的机器翻译。该方法的基本思想非常简单：首先设计模板匹配替换通用算法UAMRT，然后利用UAMRT匹配句子中的源语言模板，并将其替换为相应的目标语言模板，从而实现对句子的翻译。在结合句型分析算法和从句分析算法的基础上，利用启发式搜索机制进一步提高了句子的翻译速度和质量。速度测试表明用该方法实现的英汉翻译系统在P-IV1．7G的计算机上翻译速度每秒可以达到1300个单词左右；质量测试表明该系统的性能在开发过程中仅仅通过增加更多的模板就会变得越来越好，而且在应用时与几种商用系统相比可以达到中等水平。

Machine Translation Based on UAMRT

This paper proposes a new approach to machine translation based on UAMRT. The basic idea of this approach is very simple: it is to design UAMRT ,namely,a universal algorithm of matching and replacing templates, and translate sentence by using UAMRT to match source templates and replace them with their corresponding target templates. Combined with sentential form parsing algorithm and a clause parsing algorithm,a source sentence can be translated by heuristic scans into a number of candidate target sentences in a few translation hierarchies of classes composed of ordered source-target rules. The speed test shows that the English-Chinese translation system based on UAMRT can reach a translation speed of about 1300 words persecond on P-IV1. 7G computer; and three quality tests show that the system can not only become better and better by adding more templates without changing itself in the development,but also reach a middle accuracy in the application when comparing with commercial competing systems.