面向心理健康与改进CNN-BiLSTM的文本情感分类研究

Classification research based on text sentiment model

传统文本情感极性判断时易忽视外部信息的有效性,进而导致最终分类准确性不高。基于心理健康,构建CNN-BiLSTM-Att文本情感分类模型。首先,构建基于主题模型的BiLSTM和CNN情感分类模型;其次,将BiLSTM和CNN采用并行融合,构建CNN-BiLSTM文本情感分类模型,为提高情感极性词语分类的准确性,引入Attention机制对CNN-BiLSTM模型进行改进;最后,以Stop Words数据集对设计的模型进行实验验证。结果表明,相较于对比模型,研究构建模型性能更优越,可提升情感分类的准确率。

The effectiveness of external information is ignored in traditional text sentiment polarity judgment,which leads to the low accuracy of final classification,a text sentiment classification model based on CNN-BilLSTM-Att is constructed based on mental health.Firstly,BiLSTM and CNN sentiment classification models based on topic model are constructed respectively.Secondly,BiLSTM and CNN are combined in parallel to construct CNN-BiLSTM text sentiment classification model.In order to improve the accuracy of sentiment polarity word classification,Attention mechanism is introduced to improve CNN-BiLSTM model.Finally,the Stop Words data set is used to verify the model designed in this paper.The results show that compared with other models,the performance of the model constructed in this study is better,which can improve the accuracy of emotion classification.