摘要
以word2vec工具进行词向量运算,根据短文本语义特征,采用卷积神经网络模型提取出深度抽象特征,再对分类器进行训练来实现情感分类的目的。分析基于卷积神经网络的多维特征微博文本情感,通过F值和准确率来衡量实际分类效果,分析结果表明:相对于机器学习模型,该微博情感分析模式使情感分析和F值准确率依次增大了0.1060与0.1320。采用卷积神经网络和多维度文本特征分析方法可以有效提升微博情感分析的效果。
The word2 vec tool is selected to complete the operation of word vectors,semantic features are extracted from the short text,deep abstract features are extracted from the convolutional neural network model,and then the classifier is trained to achieve the purpose of emotion classification. By analyzing the multi-dimensional characteristics of micro-blog emotions based on convolutional neural network,the actual classification effect is measured by F value and accuracy. The analysis results show that,compared with the machine learning model,this micro-blog emotion analysis model increases the accuracy of emotion analysis and F value by 0.1060 and 0.1320 respectively. Convolutional neural network and multi-dimensional text feature analysis can effectively improve the effect of microblog emotion analysis.
作者
余鹏
田杰
YU Peng;TIAN Jie(Institute of Electric Power Science,Shenzhen Power Supply Bureau Co.,Ltd.,Shenzhen 518000)
出处
《计算机与数字工程》
2020年第9期2244-2247,共4页
Computer & Digital Engineering
关键词
情感分析
卷积神经网络
微博文本
表情字符
sentiment analysis
convolutional neural networks
Weibo text
emoticons
