期刊论文

Studying Factors of Judging Phrase Fields by Method of Principal Component Analysis

中文

计算机技术与发展

1673-629X

2008

10

116-119

基金项目：软件工程国家重点实验室开发基金资助项目（SKLSE04-018） 湖北省重点科技攻关资助项目（2003AA101C26）

本校为第一机构

对复句层次结构和层次关系进行分析和研究之前,首先要确定有标复句中分句的数量,即有标复句中的哪些字段是分句,哪些字段只是加了标点符号的句法成分（文中称之为短语字段）。结合语言学的相关理论,提取出识别短语字段的因素,并对这些因素进行主成分分析,从而得出进行识别的综合影响因素以及与原始的具体因素之间的关系。结果表明,前三个主成分所包含的信息量接近85%,已包含原有因素大部分的信息,在今后的研究中,这三个主成分将取代原来的多个变量,从而简化研究的复杂度。

Before studying the hierarchical structure and relations of the complex sentence, the number of the clauses in a complex sentence which has been tagged should have been confirmed, and which fields are the complete clauses and which are not but phrases due to the complexity of using punctuations in Chinese are the same. In the meanwhile it increases the difficulty of the later classifying the category and the layer of the tagged complex sentences. According to the linguistics theory this thesis tells the method of how to dig out and analyze the factors of finding out the phrase fields. By princ...