Hierarchical Phrase-Based Mongolian-Chinese Statistical Machine Translation

With the rapid development in fields such as economy, trade, culture and education, the communications between different peoples and regions is fastly increased. Recent years, many people came to Inner Mongolia for trade or travel. The demand of translation from Mongolian to Chinese is boosted. However, mannul translation can not meet these demands, so the machine translation became necessary. In present, state of the art machine translation methods includes phrase-based model, hierarchical phrase-based model, and syntax tree-based model. This paper aims to do research on hierarchical phrase-based machine translaiotn from Monglian to Chinese.Firstly, this paper studied hierarchical phrase-based statistical machine translation model, and then implemented a hierarchical phrase-based Mongolian-Chinese machine translaiton system. Moveover, this paper also proposed an efficient spelling correction method for Mongolian based on statistcal machine translation framework for low quality parallel corpus. Finally, experiment shows that our system could achieve higher performance than baseline system with more than1BLEU score improvement.