Sentimental Analysis for Studying and Analyzing the Spreading of COVID-19 from Twitter Data

Coronavirus, known as COVID-19, rapidly spread on a wide scale in a short time consequently. World Health Organization (WHO) classified it as a global pandemic. Social networks news becomes a valuable resource for massive amounts of data and news about the epidemic in which news is deliberating every day. Twitter is one of these networks which is a popular platform that contains rich information and currently it repre-sents a rich resource of data about COVID-19. In this research, we study and analyze the spreading of the COVID-19 epidemic based on the location and dates using datasets from Twitter. Moreover, the study has done by performing sentiment analysis and making a correlation study between confirmed cases in a set of countries and the sentiment's polarity value including negative and positive as well as a correlation between the number of confirmed cases and number of tweets per country. Also, we have experimented with several machine learning classifiers including Naive base, Support Vector Machine, and Logistic Regression as well as RoBERTa model to predict the sentiment analysis on the dataset. The experimental results show that Logistic Regression outperforms other classifiers with an accuracy of 0.86%, thus, machine learning techniques could be used to study the sentiment of tweets which gives reasonable results.