基于规则和统计的机器翻译方法歧义问题比较分析  

A Comparative Analysis of the Ambiguity Resolution of Two English-Chinese MT Approaches:RBMT and SMT

作　　者：马建军[1] 

机构地区：[1]大连理工大学外国语学院,辽宁大连116024

出　　处：《大连理工大学学报（社会科学版）》2010年第3期114-119,共6页Journal of Dalian University of Technology(Social Sciences)

基　　金：中央高校基本科研业务费专项资金资助项目(DUT10RW202)

摘　　要：文章通过对SYSTRAN和GOOGLE两个机译系统90句商务英语英汉翻译译文的比较分析,从词义歧义和句法歧义两方面,研究基于规则(RBMT)和基于统计(SMT)两种机器翻译方法的消歧处理的优与劣。研究表明,基于统计方法词义消歧强于基于规则方法,而句法消除歧义能力相当。词义歧义问题主要集中在动词多义词和名词多义词,而句法歧义问题则集中在定语、状语和从句。动词与介词搭配,名词与介词搭配作定语和状语,以及具有商务内涵的动词和名词,成为机器翻译的难点。This paper makes a comparative analysis of the ambiguity resolution of two MT approaches: Rule Based Machine Translation(RBMT) and Statistical Machine Translation(SMT),by analyzing the Chinese translation work for 90 English sentences in the domain of business,translated by SYSTRAN and GOOGLE translation systems.The results show that SMT is better than RBMT in terms of lexical ambiguity resolution,while they are the same in terms of structural ambiguity resolution.Lexical ambiguity mainly includes problems ...

关 键 词：机器翻译 词义歧义 句法歧义 RBMT SMT 

分 类 号：H085[语言文字—语言学]