摘要
针对传统情感分类算法存在的参数学习困难及分类性能较低等问题,提出了一种基于核超限学习机的中文文本情感分类方法.首先通过信息增益对训练数据进行特征选择以降低输入维数,然后通过构建基于小波核超限学习机的分类器实现对中文文本的情感分类.实验结果表明,新方法参数学习容易,且其文本情感分类性能通常优于支持向量机和朴素贝叶斯.
Aiming at the disadvantages of traditional classification algorithms for sentiment classification, such as complicated parameter learning and low classification performance, this paper proposed a novel Chinese text sentiment classification approach based on kernel extreme learning machines. First, the feature selection for training data via the information gain technology was implemented to reduce the input dimensionality. Then, a classifier based on the wavelet kernel extreme learning machine was constructed for Chinese text sentiment classification. The experimental results show that the model parameters of the proposed method are easier to learn and the Chinese text sentiment classification performance of the proposed method is usually superior to support vector machines or naive bayes.
出处
《中国计量学院学报》
2016年第2期228-233,共6页
Journal of China Jiliang University
基金
国家自然科学基金资助项目(No.61272315
11391240180)
浙江省自然科学基金资助项目(No.LY14F020041
LY15A020003)
关键词
核超限学习机
情感分类
中文文本
kernel extreme learning machine
sentiment classification
Chinese texts
