21. 基于加权词汇衔接的文档级机器翻译自动评价.pdfVIP

网络出版地址：/kcms/detail/11.2442.N1025.007.html 北京大学学报 ( 自然科学版) Acta Scientiarum Naturalium Universitatis Pekinensis doi: 10.13209/j.0479-8023.2014.004 基于加权词汇衔接的文档级机器翻译自动评价 † 贡正仙 李良友 苏州大学计算机科学与技术学院 , 苏州 215006, † E-mail: zhxgong@ 摘要 在文档词汇衔接评价 LC 方法的基础上, 提出基于权重的 LC, 即 WLC, 该方法通过在文档词图上运行 PageRank 算法获得词汇权重。根据词性信息使得 PageRank 算法偏向特定的词汇 , 并提出 PWLC 方法。实验 表明, 在文档级别上, 所提的两种方法与人工评价的相关度都优于 LC, 融合两种方法后, BLEU 和 TER 在文 档级别上的评价性能有显著的提高。 关键词 词汇衔接 ; 文档级评价; 机器翻译; 自动评价; PageRank 中图分类号 TP391 Document-Level Automatic Machine Translation Evaluation Based on Weighted Lexical Cohesion † GONG Zhengxian , LI Liangyou Department of Computer Science and Technology, Soochow University, Suzhou 215006; † E-mail: zhxgong@ Abstract Based on LC method, weighted LC (WLC) method is proposed, which assigns weights for words by PageRank algorithm running on word graph of documents . Furthermore, a new method named PWLC is also proposed, which biases PageRank algorithm to words with specific POS tags. The authors show how to combine the evaluation of lexical cohesion with other mainstream automatic evaluation metrics, in order to help these methods to evaluate translation quality at document level. Compared with LC, experiments show WLC and PWLC hav