Translating News Headlines From Chinese into English

摘要

Before approaching the book title translation in a digital library setting, we first tackle the related problem of the Machine Translation of online Chinese news headlines on sites such as sina.com.cn into English. There are four problems associated with headlines: the heavy uses of abbreviations, named entities, short ambiguous words and distinct syntactic structure. We approach these disambiguation problems by utilizing the larger context: the whole body of the story. We also use example based machine translation to further improve the translation. Compared with a rule-based baseline Chinese-English MT system Neon, the new system can produce better translations. The word error rate is reduced 26% and 71% of translations are improved for a total of 120 test headlines.