Sentiment Analysis And Visualization Of Microblog Epidemic Texts

The emergence of the novel Coronavirus Disease 2019(COVID-19)has made people eager to understand the current epidemic situation.Weibo,as a social media platform,allows netizens to learn about the real-time development of the epidemic and express their own views on the epidemic-themed platform.However,there are disadvantages to unilaterally browsing epidemic data,such as the time-consuming time-consuming statistics of user comments and the inability to obtain all netizen sentiment data intuitively.Therefore,this article proposes to use Weibo comments as the data source,display the data in a visual form,design and complete Weibo The sentiment analysis and visualization system of epidemic texts is mainly divided into three parts:(1)Based on crawler technology to achieve thematic data acquisition of Weibo epidemic situation.It mainly involves simulated user login,combined with BeautifulSoup,regular and other technologies to capture the web version of Weibo epidemic data and data filtering.The data captured in this article covers the whole year of 2020,including about 4,500 popular articles and about 500,000 popular user comments.(2)CNN-BiLSTM model based on attention mechanism and residual network performs sentiment analysis on comment data.The CNN network is used to extract data structured information,and the BiLSTM network is used to extract data time correlation and text fragment dependency information.Innovative research is carried out on the basis of this model.Dating focuses on optimizing text feature vectors and dating residuals.The network solves the problem that the number of network layers is too deep and the accuracy of training and testing drops rapidly.In the classification process,the entire text information is divided into polarity,and divided again according to the positive and negative emotions.(3)Data visualization system implementation.The combination of Vue and Node is used in front-end and back-end separation technology,and the MongoDB database is used to visualize the data through the ECharts framework and in the form of charts.This model is compared with LSTM,BiLSTM and other models.The experimental results show that the prediction accuracy and recall rate have been significantly improved under different data sets.At the same time,through natural language processing combined with Web development technology and epidemic data,users can have a more comprehensive understanding The epidemic incident and the emotional bias behind the incident will also provide a reference for epidemic prevention and control.