摘要
提出一种具有特征级别的领域特征集合的情感资源挖掘方法,将基于HowNet词典的分类法构建的情感特征与基于机器学习的特征分类方法中的无内容特征以及领域特征相融合,并将该集合放入支持向量机中进行情感分类实验,实验结果表明,使用抽取模式以及多特征融合的分类方法,可增强中文情感分类效果,验证两种分类方法综合研究的正确性与有效性,弥补目前特征级别的中文情感分类研究的不足。
In order to establish a verified sentiment collection with integrated features, the paper bases on HowNet dictionary ap- proach, and formes a strong sentiment collection with the exiting no-content and field features in the machine learning approach. The sentiment collection is put into a vector machine to do sentiment classification. The result shows that the collection improves the Chi-nese sentiment effect via an extracting method and integrated features. The collection also verifies the feasibility and significance of the two approaches, and makes up the present lack of Chinese sentiment classification research.
出处
《图书情报工作》
CSSCI
北大核心
2012年第21期109-113,61,共6页
Library and Information Service
关键词
情感分类
情感特征
无内容特征
领域特征
sentiment classification ,sentiment feature, no-content feature ,field feature
