摘要
针对"把"字句在机器翻译中的困难,探索一种规则和统计相结合的"把"字句的自动释义和句式变换的方法。具体的计算步骤为:(1)根据"把"字句与其他句式的变换关系,将"把"字句分为不同的小类,并总结出每一小类的句法语义特征,得到"把"字句的语言模型;(2)选取北大中文树库中的"把"字句作为语料,并标注上每一小类句式的句法语义特征,从而得到富含句法语义信息的标注文本;(3)在此基础上,分别用组块分析的方法和完全句法分析的方法来对"把"字句进行自动识别;(4)再利用判别式机器学习的方法来对"把"字句进行自动分类。在识别结果和分类结果的基础上,根据释义模板和变换模板得到了一个"把"字句的自动释义与句式变换程序。
In order to enhance the accuracy of the automatic interpretation of the b^-construction, this paper adopts a com- putational oriented and cognitive based schema, and attempts to have an automatic analysis of the semantic interpretations and syntactic alternations of the bS-construction. The paper firstly classifies the b^-construction into different subtypes based on the alternations of the b^-construction with other constructions and summarizes the syntactic and semantic features of each subtype. Then, it builds up a language model of the ba-construction, with an annotated text filled with the syntactic and semantic features of the ba-construction. Further, it automatically identifies the ba-construction based on the chunking and parsing methods, and automatically classify them based on machine learning. Finally, the paper designs an automatic interpretation and alternation system of the bfi-construction.
出处
《计算机工程与应用》
CSCD
北大核心
2015年第19期129-137,共9页
Computer Engineering and Applications
基金
国家社科基金重大招标项目(No.12&ZD175
No.12&ZD227)
关键词
“把”字句
变换分析
框架识别
自动分类
自动释义
the ba-construction
pattern alternation
frame identification
automatic classification
automatic interpretation
