基于HMSVM模型的中文浅层句法分析  

Chinese Shallow Parsing Based on HMSVM Model

作　　者：王仲华[1] 卢娇丽[2] 付继宗[3] 

机构地区：[1]太原师范学院数学系,太原030012 [2]山西大学现代教育技术中心,太原030006 [3]北方自动控制技术研究所,太原030006

出　　处：《电脑开发与应用》2013年第2期30-32,共3页Computer Development & Applications

摘　　要：隐马尔科夫支持向量机(HMSVM)是一种新颖的结构化支持向量机模型,该模型在序列标注学习任务比如英文组块分析中的有效性已经被证明,将该模型用于中文浅层句法分析识别任务,实验结果表明,该模型获得了较好的准确率和召回率。Hidden Markov Support Vector Machines is a novel structural SVMs model. Its efficiency has been proved in label sequence learning task such as English text chunking. In this paper, we treat Chinese chunk recognition as a label sequence learning problem. After giving the definition of Chinese shallow parsing, we apply HMSVM to solve Chinese shallow parsing problem. The results of experiment show that it achieves a better precision and recall.

关 键 词：浅层句法 隐马尔科夫支持向量机 序列标注 边界识别 

分 类 号：TP391.1[自动化与计算机技术—计算机应用技术]