融合直推式学习和语义理解的词语倾向性识别  

Identifying of Word Sentiment Orientation of Transductive Learning and Semantic Comprehension

作　　者：闻彬[1] 饶彬[1] 赵君喆[1] 焦翠珍[1] 戴文华[1] 

机构地区：[1]湖北科技学院计算机科学与技术学院,湖北咸宁437100

出　　处：《计算机技术与发展》2016年第1期74-77,共4页Computer Technology and Development

基　　金：国家自然科学基金面上项目(61373108);湖北省教育厅科研项目(Q20112809;B20082803);湖北省教育厅人文社会科学研究项目(13g389)

摘　　要：目前词语情感倾向性识别研究主要分为机器学习和语义理解,机器学习不能很好地识别通用领域词语,语义理解又存在准确率和召回率不够高的问题,因此文中提出了一种融合直推式学习和语义理解的词语倾向性识别方法。首先对HowNet知识库体系进行改进,在已有的四种义原的基础上,提出第五义原—情感义原;然后将第五义原手工融入到HowNet知识库中,再在此基础上提出词语情感相似度计算方法计算词语的情感值;最后将该方法融合直推式学习以判定词语情感倾向性。通过实验结果表明,与支持向量机和原语义理解方法相比,该方法在识别情感词上取得了较好的效果。At present,the research on word sentiment orientation identification is mainly divided into machine learning and semantic comprehension,but machine learning cannot handle general field words effectively,semantic comprehension also cannot get high scores at precision and recall,therefore,a newfusion method between transductive learning and semantic comprehension for judging word polarity was put forward in this paper. Firstly the HowNet knowledge base system is improved,on the basis of four primitive,the fifth primitive—sentimental primitive was proposed,which was integrated into HowNet manually,on the basis of this,then a newword sentimental similarity calculation method was proposed to compute word＇s sentimental value. At last,combine this way with transductive learning for identifying word＇s sentimental orientation. The performance of experiment shows that compared with SVM or traditional semantic comprehension,it can get better results.

关 键 词：词语倾向性识别 机器学习 语义理解 意见挖掘 情感义原 HOWNET 

分 类 号：TP391.1[自动化与计算机技术—计算机应用技术]