一种基于LSTM和ResNet网络的情感极性分析方法

Sentiment Polarity Classification Based on ResNet and LSTM

人们习惯在社交平台分享生活、发表看法、发泄情感,由于数据量大且易于获取,社交平台文本数据已被广泛用于网络用户情感极性分析。因此,文本情感分析方法也经常用来对网络舆情进行研判和预测。传统对文本情感极性分析的方法没有应用深度学习等成果技术,使得情感分类结果的准确性不高。提出一种ResNet残差网络改进的LSTM长短时间序列分析方法。实验结果表明,与支持向量机、朴素贝叶斯等传统分类器相比,基于改进的ResNet与LSTM的文本情感极性分类方法在分类精度上有一定提升;与LSTM、循环神经网络等深度学习方法相比,该方法在保证运行效率的前提下能获得更高的分类精度。所提方法能够用来对社交平台的文本情感进行情感极性分类和预测。

People are used to sharing their lives,expressing opinions,and venting emotions on social platforms.Due to the large amount of data and easy access,text data on social platforms have been widely used to analyze the emotional polarity of Internet users.There-fore,text sentiment analysis method is often used to judge and predict online public opinions.The traditional method of analyzing text e-motion polarity does not adopt the deep learning technology,so the accuracy of the result of emotion classification is not high.An im-proved LSTM short and long time series analysis method based on ResNet residual network is proposed.Experimental results show that compared with traditional classifiers such as support vector machine and Naive Bayes,the classification accuracy of text emotion polarity classification method based on improved ResNet and LSTM is improved to some extent.Compared with deep learning methods such as LSTM and recurrent neural network,the proposed method can obtain higher classification accuracy under the premise of ensuring opera-tion efficiency.This method can be utilized to classify and predict the emotional polarity of text emotions on social platforms.