【摘要】 词对齐被广泛的用于基于短语的统计机器翻译中,词对齐效果的好坏直接影响了机器翻译的质量。提出将隐马尔科夫模型用于汉维词对齐时,由于汉维双语标记的数据量比较大而且标记数据也还没有公开,导致汉维词对齐的质量较差,也没有办法进行评价,提出采用基于词典的方法进行对齐评价,实现汉维双语词典的构建系统,实验表明,该方法的效果较好,并同时构建汉维双语语料库。更多还原

【Abstract】 Word alignment is widely used in statistical machine translation phrase based on phrase. The effect of word alignment directly affects the quality of machine translation. Puts forward using a hidden Markov model for Chinese-Uyghur word alignment, because of the large amount of bilingual marker data and the lack of labeled data, resulting in poor quality of Chinese Uyghur word alignment, there is no way to evaluate. Puts forward the evaluation method based on the alignment dictionary and constructs a bilingual dictionary system. The experiment shows that the effect is good and the Chinese Uighur bilingual corpus is constructed.更多还原