引用本文:周 虎,于 跃,贾媛媛,赵文龙.基于深度LSTM神经网络的在线消费评论情感分类研究[J].中华医学图书情报杂志,2018,27(5):23-29.
基于深度LSTM神经网络的在线消费评论情感分类研究
Deep long-short term memory neural network-based sentiment classification of online reviews on consumption
DOI:10.3969/j.issn.1671-3982.2018.05.005
中文关键词:  在线评论  情感分析  神经网络  支持向量机  深度学习
英文关键词:Online review  Sentiment analysis  Neural network  Vector-supporting machine  Deep learning
基金项目:国家自然科学基金“基于深度卷积生成对抗网络的3D-MRI图像超分辨率重建研究”(61702064)和中国博士后科学基金“基于稀疏表示的4D-MRI图像超分辨率重建研究”(2017M623303XB)的研究成果之一
作者单位
周 虎 重庆医科大学医学信息学院重庆 400016 
于 跃 重庆医科大学医学信息学院重庆 400016 
贾媛媛 重庆医科大学医学信息学院重庆 400016 
赵文龙 重庆医科大学医学信息学院重庆 400016 
摘要点击次数: 28
全文下载次数: 10
中文摘要:
      目的:探索深度神经网络在情感分类方面的应用。方法:采用长短型记忆神经网络构造在线消费评论情感的分类器,对收集的消费评论进行二分类的情感分析。结果:LSTM模型的准确率为89.29%,优于实验对照SVM模型的86.10%,深度神经网络模型在本文的情感分类中的准确率较高。结论:使用深度神经网络对消费评论的情感进行分类,可减少人工特征的干预,提高在线消费评论情感分类的效率。
英文摘要:
      Objective To study the application of deep long-short term memory (LSTM) neural network in sentiment classification.Methods A sentiment classifier for online reviews on consumption was made using the LSTM neural network. The binary catalogued sentiment of recorded online reviews on consumption was analyzed. Results The accuracy of LSTM model was higher than that of SVM model for sentiment classification (89.29% vs 86.10%). A rather high accuracy of deep neural network was achieved for sentiment classification in this study. Conclusion Deep neural network-based sentiment classification of online reviews on consumption can reduce the intervention of artificial features and improve the sentiment classification efficacy of online reviews on consumption.
查看全文  查看/发表评论  下载PDF阅读器
关闭
手机扫一扫看