汉语词汇主观性提取研究  

Study on subjectivity extraction of Chinese vocabulary

作　　者：张靖[1] 金浩[2] 

机构地区：[1]攀枝花学院网络中心,四川攀枝花617000 [2]南京大学计算机科学与技术系,江苏南京210093

出　　处：《计算机工程与设计》2010年第4期915-917,F0003,共4页Computer Engineering and Design

基　　金：四川省教育厅科研基金项目(07ZB155);攀枝花市科技基金项目(2009CY-R-13)

摘　　要：为了在情感分析中有效实现汉语词汇主观性提取,提高提取准确率,分析和研究了汉语词汇主观性提取方法,建立了基于二元语法依赖关系的语言模型,由此定义了主观性提取模式,设计了自适应主观性自举算法,利用情感倾向特征和主观性强度特征集合进行特征选择和建模,提高了词汇主观性判断的准确率,使用机器学习方法构造分类器实行主观性判断。实践结果表明,该方法能有效的实现了词汇主观性获取,性能得到了提高,提取准确率达到88.64%。To realize efficiently the subjectivity extraction of the Chinese vocabulary in the sentiment analysis and to advance the accuracy of the sentiment analysis,the means of extracting the subjectivity of the Chinese vocabulary is analyzed and researched,the linguistic model is created based on the bi-gram dependency,thus the subjectivity extraction pattern is defined,and the adaptive bootstrap algorithm is designed.The sentiment polarity feature and subjective intensity set are used for the feature selection and modeling to improve the accuracy of the subjectivity extraction.The classifier constructed by using the machine learning,is used to judge the vocabulary subjectivity.It is proved that the vocabulary subjectivity extraction is available,and the performance is improved,and the accuracy of the subjectivity extraction reaches 88.64%.

关 键 词：自适应 自举 提取 特征选择 模型 主观性 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]