摘要
在实用机器翻译系统的研究开发中 ,复杂长句的翻译处理是其面临的一个主要难题。本文提出一种多语种通用的基于多策略分析的复杂长句翻译处理算法 ,该算法通过基于实例模式匹配和规则分析相结合的方法 ,综合利用源语言句子中多种相关的语言特征 ,包括语法语义特征、句子长度、标点符号、功能词以及上下文语境条件等对复杂长句进行切分简化处理和译文的复合生成。另一方面 ,通过对不同语种设计相同的知识表示形式 。
The processing of complex long sentences is a difficult problem in the implementation of a practical machine translation system.In this paper,a processing approach to the complex long sentence based on multi strategy is proposed,in which the rule based analysis and case based consistent matching are combined by using various linguistic features,including sentence length,punctuation,functional words and contextual condition and so on,which is used to segment the complex sentence into several simple sentences and to compose the translation text of these simple sentences together.In addition,by designing the uniform formalism of the knowledge representation,it is possible for this approach to be suitable to multi lingual translation system.
出处
《中文信息学报》
CSCD
北大核心
2002年第3期1-7,共7页
Journal of Chinese Information Processing
基金
国家自然科学基金
杰出青年基金(6 992 5 10 2 )支持