Sentiment Analysis Model of Specific Target Text Based on Convolutional Neural Network
In the process of emotion analysis of specific target text,text sequence classification is limited by the labeling method,resulting in low accuracy and recall of analysis results,so a model of emotion a-nalysis of specific target text based on convolution neural network is constructed.Specific target text se-quences are analysed based on emotional differences,and the text feature matrix is used as input data to construct a convolutional neural network language model in the input layer,concatenating it into a part of speech sequence matrix.Segmented pooling captures different key features of text sequences and classi-fies and processes the extracted feature vectors.A dropout mechanism is added to complete sentiment classification of specific target texts,determine the importance information of each word in the text,and achieve sentiment analysis of specific target texts.The experimental results show that the accuracy of the model is higher than 84％,and the maximum recall rate is 87％,which can effectively achieve the emo-tional analysis of specific target text.
国家科技期刊平台
NSTL
万方数据
