摘要
针对长句子引起句法分析性能下降的问题,本文提出了一种基于SVM的句子片段划分方法:先根据语法结构将句子划分为多个片段,识别出每个片段的类别;然后根据片段的类别将句子分割为几个部分,每个部分作为句法分析的基本单元;最后将句法分析之后的各个部分进行合并,形成完整的分析结果.该方法减小了句法分析的复杂度,提高了分析的准确率.
Aimed at the decreased performance of syntactic parsing caused by long sentence, this paper presents a method of identifying the segments based on the SVM classifier to solve this problem. In this method, a sentence is firstly divided into different segments, each of which is assigned a label to indicate its syntactic type. Then the sentence is parsed based on the segments. Finally, all the segments are linked together through the dependency relations and the parsing of the whole dependency tree is completed. Experiments show that the identification of segments decreases the complexity of parsing and improves the accuracy of Chinese dependency parsing.
出处
《哈尔滨工业大学学报》
EI
CAS
CSCD
北大核心
2009年第5期52-55,共4页
Journal of Harbin Institute of Technology
基金
国家自然科学基金资助项目(60575042
60675034)
