2019年基于规则的英汉翻译技术报告.pptVIP

基于规则的英汉翻译技术报告;1 背景情况;2 系统结构 ;2.1 搜索引擎;如：“People have a variety of emotion,” argues Harvard, “Some people handle anger well but can‘t handle fear. So each emotion has to be viewed differently.”需要还原成Harvard argues ,“People have a variety of emotion,Some people handle anger well but can’t handle fear. So each emotion has to be viewed differently.”。型如U.S.、Mr.、www.等带.标点的缩写从形式上会对分句产成干扰，对缩写的识别可以消除这一问题。 搜索引擎完成了搜索句中每个单词的词性，释义，合成词识别、数词识别，时间年月识别，固定表达等识别，短语、模式、规则等的搜索匹配。搜索引擎是句子翻译的起点，同时贯穿释义选择的整个过程。搜索引擎首先建立每个单词的词性、属性、文体属性、语义属性等信息，为成分识别和释义选择提供服务，在释义选择的过程中搜索引擎需要针对一些短语、模式或规则进行关联搜索匹配，提供翻译的服务。 ;2.2成分识别;2.2.1底层短语识别;The boy’s looking at the tall tree.需要还原成The boy is looking at the tall tree. He’s finished the work.需要还原成He has finished the work. The boy’s book is lost.需要将boy’s识别成所属格。 Hed finished.需要还原成He had finished. Id like to have a cup of tea with you on this Sunday.需要还原成I would like to have a cup of tea with you on this Sunday. ;He found a way to take an udder cell and make it grow into a new cloned lamb.需要还原成He found a way to take an udder cell and make it to grow into a new cloned lamb. 一般疑问句，可以先还原成陈述句，并给出翻译模式，句尾加“吗”。 Do you think the proposal is acceptable?可以还原成you think the proposal is acceptable? Is the station clock accurate?可以还原成the station clock is accurate? 常见需要转换的句型有疑问句(包括一般疑问句、反义疑问句、特殊疑问句、倒装句、强调句等)。;为了降低单词多词性的干扰和利用不同短语类型的分类识别，短语识别采用从句子的最后一个单词向前推进，从而识别短语的类别。如果短语的最后一个的单词具有唯一的词性，短语的类型相对确定；如果短语的最后一个的单词词性不确定，需要结合前后环境进行多词性单词的识别，从??确定短语最后一个单词的词性。 对于短语前面修饰成分的识别，合并成一个短语。确定短语最后一个单词的词性后，根据短语最后一个单词的词性不同，分为不同类型短语，进行短语前面修饰成分的识别，最终将相关联的单词组合成一个短语。 如An advantage of using the solar energy is that it wont create any pollution.短语识别的结果为：An advantage是名词短语、of using是介词短语、the solar energy是名词短语、is是谓语短语、that是从句引导词、it是代词短语、will not create是谓语短语、any pollution是名词短语、.是标点。;2.2.3一级成分识别;短语识别的基础上进行短语的合并;短语合并基础上识别合并后短语的一级成分; An advantage 是名词短语---主语 ---An advantage---是名词短语 ---of using---是介词短语 ------of using---是介词短语 ------the solar energy---是名词短语 ---------the solar energ