实例统计翻译混合策略的汉民病历翻译的研究

A Study on Chinese Minority Medical Document Translation Based on Hybrid-strategy

针对民族地区不同语言民族患者看病求医中汉文处方理解难、病历及药物信息不易读懂等问题,本文研讨基于实例及基于短语的统计机器翻译混合策略的汉民电子病历、门诊处方及药物信息等医疗卫生用文机器翻译方法,试图实现高精度保质量的医药文翻译系统,提升民族地区公共卫生服务水平.本次汉-维门诊处方文本翻译试验结果BLUE值显示,本文提出方法比单一翻译方法(如实例或者统计翻译)显著提高了译文质量.

In minority ethnic groups living areas such as Xinjiang and Inner Mongolia, there are usually communication problems between the patients from the minority groups and the medical workers due to the different languages they daily used. Also for patients from the minority groups, It is not easy to understand the prescribes, medical reports, the names of medicines, which are mostly given in Chinese. It is very important that the understanding of the disease name and symptoms correctly, and use medicine safely for both the medical workers and patients.Pointed above problems, this research will challenge to develop a machine translation system combined a multi-lingual medical information parallel sentence set and phrase-based statistical machine translation（PBSMT）. From results of experiments, we obtained a BLEU score of 0.3132 in the medical Chinese-Uyghur task using our proposed method. In contrast, we had another BLEU score of 0.2821 in the medical ChineseUyghur task using only the PBSMT. This means that the proposed machine translation approach is very fast and effective for converting the medical documents written by Chinese or Minority.