Emotion recognition and analysis of netizens based on micro-blog during covid-19 epidemic

Abstract

The research is about emotion recognition and analysis based on Micro-blog short text. Emotion recognition is an important field of text classification in Natural Language Processing. The data of this research comes from Micro-blog 100K record related to COVID-19 theme collected by Data fountain platform, the data are manually labeled, and the emotional tendencies of the text are negative, positive and neutral. The empirical part adopts dictionary emotion recognition method and machine learning emotion recognition respectively. The algorithms used include support vector machine and naive Bayes based on TFIDF, support vector machine and LSTM based on wod2vec. The five results are compared. Combined with statistical analysis methods, the emotions of netizens in the early stage of the epidemic are analyzed for public opinion. This research uses machine learning algorithm combined with statistical analysis to analyze current events in real time. It will be of great significance for the introduction and implementation of national policies.