Research On Sentiment Analysis Method And Application Based On Short Text Classification

In recent years,with the rapid development of Internet technology,the Internet text data shows explosive growth.How to process and analyze the massive text data quickly,efficiently and accurately,and obtain the emotional tendency of users has become crucial.Emotion analysis based on deep learning has the characteristics of low cost,high efficiency and strong learning and expression ability,and has become one of the research hotspots in the field of emotion analysis.However,the existing text sentiment analysis methods have some problems such as poor feature extraction ability and poor sentiment classification performance.In order to effectively solve the above problems,the paper studies the sentiment analysis method based on graph convolutional neural network.By combining the graph convolutional neural network model with BERT pretraining model,Res Net residual structure and bidirectional long short-term memory network model,the relationship between short text context is mined,and better word text vector representation is learned to improve the performance of the model emotion analysis.The main research contents are as follows:(1)Aiming at the problem that the performance of sentiment analysis in the current field of sentiment analysis is not ideal,a sentiment analysis method based on BERT-GCNRes Net is proposed.This model can obtain better text vector representation and richer short text context connections,so as to enhance the feature extraction ability of the algorithm.Experimental results show that BERT-GCN-Res Net sentiment analysis algorithm can extract more semantic information from short texts,and improve the performance of sentiment analysis model.(2)On the basis of the above research,in order to further improve the performance of the sentiment analysis model,obtain more semantic characteristic information,and better obtain the context information of the short text context,the emotion analysis model based on BERT-Bi LSTM-GCN is proposed.This model combines the advantages of BERT module and GCN module,and makes up for the deficiency of short text context information extracted by BERT-GCN-Res Net sentiment analysis model.Because the long short-term memory network has a strong ability to acquire text context information,the forward and backward long short-term memory network modules are introduced into the model to further integrate the context information extracted from the forward and backward networks,which can better obtain the context information of sentiment short texts.The experimental results further prove that BERT-Bi LSTM-GCN sentiment analysis model can effectively improve the accuracy of text sentiment classification and has more ideal sentiment analysis performance.