Sentiment Analysis Model Of Chinese Microblog Based On Deep Learning

With the continuous development of network information technology,text mining and processing are gradually being carried out by computers,and sentiment analysis is such a technology.As a natural language processing method,sentiment analysis can effectively analyze the subjective intentions contained in text information,and has a wide range of applications in product satisfaction surveys,microblog attitude discrimination,etc.There are three main types of microblog sentiment analysis technologies:statistics-based,machine-learning-based and deep-learning-based.The main research object of this paper is a Chinese microblog sentiment analysis model based on deep learning.Aiming at the deep learning model for microblog sentiment analysis,this paper mainly conducts research from the following two aspects:1.Chinese microblog sentiment analysis model based on TextCNN and BiLSTM:For traditional deep learning network models(such as CNN and LSTM),the feature extraction is relatively simple,and it cannot effectively extract different types of features in the microblog text.This paper proposes a Chinese microblog sentiment analysis model based on TextCNN and BiLSTM.The model not only considers the information contained in the text itself,but also considers the emotional information carried by the emoji,and converts the emoji to text processing.At the same time,combined with the feature that CNN can effectively extract local features,a TextCNN model adjusted for text is used to extract the spatial features of text.Then,the BiLSTM model is used to extract the time series features contained in the text,and finally a comprehensive microblog sentiment analysis model is formed.Experiments show that the model can effectively extract the time series feature information and the spatial feature information of phrases in the text,and the accuracy and F1 value have been improved.2.Chinese microblog sentiment analysis model with enhanced dual attention:Aiming at the problem that the former model cannot distinguish the importance of different information in microblogs,and cannot make corresponding importance adjustments for specific emotional words or phrases,a Chinese microblog sentiment analysis model based on a two-layer attention mechanism is further proposed.The model utilizes the feature weight assignment technology of the attention mechanism,and can not only learn the long-distance interdependence between microblog texts,but also assign weights to the extracted features at different levels(word level and sentence level),thereby strengthening the The model’s ability to extract key information and achieving a better classification effect.Through comparative experiments,it is found that the Chinese microblog sentiment analysis model with enhanced dual attention has improved accuracy and F1 value compared with the previous model,which proves the effectiveness of the model.