Deep Learning-Based COVID-19 Twitter Analysis

ABSTRACT

Since the outbreak of COVID-19, the pandemic has impacted billions of people's lives around the world. Social media, such as Twitter, has been one of the major platforms where people express their emotions and thoughts about the unprecedented pandemic. In this paper, we perform Twitter sentiment analysis to gain insights into the development of Twitter users' sentiments during the period from February 1 to December 31, 2020. We use Long Short-term Memory (LSTM), a deep learning-based Natural Language Processing (NLP) method, to detect multiple sentiments out of eleven kinds. We also picked a number of topics of interest, such as social justice, mental health, vaccines, and misinformation, and conducted theme-specific sentiment analysis. In order to delve deeper into the meaning behind the sentiment trends, we used the Latent Dirichlet Allocation (LDA) algorithm to perform theme-specific topic modeling, which reveals interesting results. © 2021 ACM.