语义关系下的英语长句机器翻译算法优化  

Optimized machine translation algorithm for english long sentences based on semantic relation

作　　者：王晶[1] 赵彩[1] WANG Jing;ZHAO Cai(City College of Xi’an Jiaotong University,Xi’an 710018,China)

机构地区：[1]西安交通大学城市学院,西安710018

出　　处：《信息技术》2021年第8期102-105,111,共5页Information Technology

摘　　要：专业文献具有文本复杂度高、分布领域广、新词多的特点,传统机器翻译难以获得精确的翻译结果。提出一种句子切分算法,即将长句切分为多个小句后翻译。算法使用概念层次网络(HNC)理论对特征语义块进行预处理,基于语义特征和逻辑概念定义切分小句,利用规则的翻译系统进行小句翻译,通过顺序调整和组合得到全句的翻译结果。实例验证结果表明:融合了切分算法的翻译系统BLEU值达到了0.1886,相较于未融合系统提升了30%,表明该翻译算法能有效提升对专业文献复杂句式的翻译效果。Professional literature has the characteristics of high text complexity,wide distribution and many new words,which is difficult for traditional machine translation to obtain accurate translation results.In this paper,a sentence segmentation algorithm is proposed,which divides long sentence segmentation into multiple clauses before translating.By using the concept layer side network(HNC)theory to preprocess the feature semantic block,defining the segmentation of the clause according to the semantic feature and logical concept.Then,the segmentation is translated by translation machine based on rules,and the final result would obtained by adjusting and combining the segments.Example verification results show that the BLEU value of the translation system with segmentation algorithm reachs 0.1886,which is 30%higher than that of the unintegrated system,which indicates that the translation algorithm can effectively improve the translation effect of complex sentence patterns in professional literature.

关 键 词：机器翻译 切分算法 复杂长句 BLEU值 

分 类 号：TP392[自动化与计算机技术—计算机应用技术]