Sentiment Analysis of Twitter Corpus related to COVID-19 induced Lockdown

Saini, K.; Vishwakarma, D. K.; Dhiman, C.

During the global COVID-19 outbreak, social media has been the biggest platform for the rapid spread of both accurate and inaccurate information. This infodemic has given rise to a plethora of misinformation manifesting in the form of mass fear and anxiety. According to a study conducted by the Oxford Researchers, 59 percent of the inaccurate information associated with Covid pandemic that has been invalidated by the fact checker remain on Twitter without any prior warning1. This paper analyzes the psychometric impact of COVID-19 infodemic using Stacked LSTM model and Stanford's pretrained GloVe embeddings. In this, sentiment score ranging from-1 to +1 was computed for more than 6.7 tweets, where-1 denotes purely negative sentiment and +1 represents positive sentiment. This research provides insights into Corona virus psychometric progression, and outlines the proposed methodology, its implications, limitations and opportunities. © 2021 IEEE.