Sentiment Analysis and Classification of COVID-19 Tweets

Abstract

The ruinous COVID-19 (Coronavirus disease) pandemic is the paramount issue prevalent in our world today. Social media serves as a common platform, where thoughts, experiences and essential information pertaining to COVID-19 could be shared. The primary objective of this project is to analyze using NLP (Natural Language Processing) techniques, the sentiments of users across various countries based on tweets posted during the pandemic, which could be beneficial for healthcare and government organizations to assess and address the needs of individuals and formulate policies accordingly. The sentiments of tweets are determined to be positive or negative based on analysis results obtained. Unlike existing research, a proper comparative sentiment analysis of COVID-19 related tweets has been performed to obtain conclusions regarding suitability of sentiment classification models and their respective accuracies. According to the proposed approach in this paper, a convolutional neural network and recurrent neural network is constructed for sentiment analysis based on text, which will help identify the growth in fear sentiment and negative sentiment with assured higher accuracy.