![](data:image/png;base64,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)
144 Chapter 5. 基于词的机器翻译建模 肖桐 朱靖波
等人提出了一个观点
[10]
:在翻译一个句子时,可以把其中的每个单词翻译成对应的
目标语言单词,然后调整这些目标语言单词的顺序,最后得到整个句子的翻译结果,
而这个过程可以用统计模型来描述。尽管在人看来使用两个语言之间对应的单词进
行翻译是很自然的事,但是对于计算机来说可是向前迈出了一大步。
先来看一个例子。图 5.1展示了一个汉语翻译到英语的例子。首先,可以把源语言
句子中的单词“我”、“对”、“你”、“感到”和“满意”分别翻译为“I”、“with”、“you”、“am”
和“satisfied”,然后调整单词的顺序,比如,“am”放在译文的第 2 个位置,“you”
应该放在最后的位置等等,最后得到译文“I am satisfied with you”。
我
对
你
感到
满意
I
am
satisfied with
you
图 5.1 汉语到英语的翻译实例及两种语言单词之间的对应关系
上面的例子反映了人在做翻译时所使用的一些知识:首先,两种语言单词的顺
序可能不一致,而且译文需要符合目标语的习惯,这也就是常说的翻译的流畅度;其
次,源语言单词需要准确地被翻译出来,也就是常说的翻译的准确性问题和充分性
问题。为了达到以上目的,传统观点认为翻译过程需要包含三个步骤
[17]
:
• 分析:将源语言句子表示为适合机器翻译的结构。在基于词的翻译模型中,处
理单元是单词,因此在这里也可以简单地将分析理解为分词
1
。
• 转换:把源语言句子中的每个单词翻译成目标语言单词。
• 生成:基于转换的结果,将目标语译文变成通顺且合乎语法的句子。
源文:我对你感到满意
分析
转换
生成
我
对
你
感到
满意
I
am
satisfied with
you
译文:I am satisfied with you
图 5.2 翻译过程中的分析、转换和生成
1
在后续章节中会看到,分析也包括对句子深层次结构的生成,但是这里为了突出基于单词的概念,
因此把问题简化为最简单的情况。