Research On Emotional Analysis Of COVID-19 Micro-blog Based On Long Short-term Memory

With the boom development of online social media and the promotion of living standards,online text information is also developing rapidly.Microblog is the most popular blog service in China at present.It produces a lot of data at any time.There are a series of information about the feelings and attitudes of people in these data.If we can apply emotion analysis technology to a large number of text information generated by microblog,we can supervise and guide the emotional trend of netizens,so that the relevant departments can take timely measures to solve the problem.Text topic analysis and public opinion analysis are also based on emotion analysis.Exploring the emotional value of data helps to improve the network public opinion detection system.The main research contents of this thesis are as follows:(1)Data collection and pretreatment of COVID-19 reviews.Web crawler is used to collect the COVID-11 micro blog data.The Selenium framework can better simulate the behavior of human operation of the browser as well as the processing of dynamic loading data and encrypted data to avoid the anti-crawler mechanism of Micro-blog.Finally,the web page content is analyzed and extracted and stored in the My SQL database.Data on preconditioning including category marking,data cleaning,words dividing,stop words processing and other operations,so as to obtain the data set can be used to structure the model.(2)Microblog text sentiment analysis model training.In order to solve the problems of gradient disappearance and gradient explosion in the long-term training process,this thesis improves the emotion classification method based on the long-term and short-term memory network model.By constantly adjusting and changing the epoch and dropout parameters,we can find the most suitable parameters for this model.The experiment analyzed the effect of sentiment analysis of deep learning models including Recurrent Neural Network,Long and Short Term Memory neural network and Gate Recurrent Unit.The results showed that LSTM’s classification of COVID-19 micro-blog reviews had two long-term effects,and improved the ability of long-term dependence,better retaining important features and predicting the text more accurately.The emotional tendency of this project is to build a better model for the follow-up emotional analysis application system.(3)Design and implementation of sentiment analysis system for COVID-19 microblogs.Based on the micro-blog text sentiment analysis model,a COVID-19 micro-blog text sentiment analysis system based on long and short term memory network is designed and implemented according to its needs analysis.Through the front-end interface of the system,a series of visual analysis methods for new crown disease are provided to users.Based on the long short-term memory network,this thesis conducted in-depth analysis and research on the sentiment analysis of the COVID-19 text on Micro-blog,aiming to make effective monitoring and decision-making before and after the outbreak of online public opinion,so as to improve the ability to solve the unknown and unknown risks related to public opinion.