LSTM-RNN Based Sentiment Analysis to Monitor COVID-19 Opinions using Social Media Data

Abstract

Lately people are using social media for sharing their thoughts, insights about different topics and issues. The main aim of social media is to connect users and update their statuses/thoughts. One of the most used online social networking sites to share information is Twitter with roughly 330 million uses globally and 48.35 million US users where they share their opinions and thoughts. Recently, the world faced a serious pandemic, Corona virus disease (COVID-19) outbreak and the World Health Organization (WHO) declared the virus as a global health emergency. The COVID-19 started in late December of 2019 in Wuhan City, Hubei Province, China. Around the world during this time, individuals use social media to share their opinions about the pandemic. Because of the lack of information about the virus, people switched to micro-blogging platforms such as Twitter. In this study, we utilize natural language processing (NLP) techniques for opinion mining to extract negative and positive sentiments/tweets on COVID-19. We investigate NLP based sentiment analysis using Recurrent Neural Network (RNN) model with Long-Short Term Memory networks (LSTMs). Predicted sentiment using LSTM-RNN, which gives high accuracy, can be used to educate people about the virus.