节点文献
基于规则的名词短语预调序
Noun Phrase Pre-reordering based on Rules
【Author】 Mou Xiaofeng,Xun Endong Center of Language Information Processing,Beijing Language and Culture University,Beijing,100083
【机构】 北京语言大学语言信息处理研究所;
【摘要】 短语预调序是提高机器翻译效果的有效手段。本文通过对英汉翻译中名词短语结构的分析,总结得到名词短语调序规则,通过将这些规则进行短语预调序,提高了最终的翻译效果。为了缓解专家规则覆盖范围太大的弊端,本文从大规模双语语料中自动获取实例规则。通过使用专家规则与实例规则进行短语预调序,进一步提高了英汉翻译的效果。
【Abstract】 Phrase pre-reordering is a effective means to improve machine translation performance.We summarized some reordering rules of noun phrase through analyzing noun phrase structures in English-Chinese translation.And the experiment shows that translation is improved through the use of these rules.In spite of such improvement,expert rules are somewhat coarse.To deal with such defects,lots of reordering examples is collected.And the experiment shows that translation performance is improved,compared with that in previous experiment.
- 【会议录名称】 第五届全国青年计算语言学研讨会论文集
- 【会议名称】第五届全国青年计算语言学研讨会
- 【会议时间】2010-10-11
- 【会议地点】中国湖北武汉
- 【分类号】TP391.2
- 【主办单位】中国中文信息学会