Beyond Fear and Anger: A Global Analysis of Emotional Response to COVID-19 News on Twitter Using Deep Learning

13 Pages Posted: 21 Mar 2023

See all articles by Francisco Braulio Oliveira

Francisco Braulio Oliveira

University of Sao Paulo

Davoud Mougouei

Deakin University

Amanul Haque

North Carolina State University

Jaime Simão Sichman

University of Sao Paulo

Hoa Khanh Dam

University of Wollongong

Simon Evans

University of Surrey - Department of Psychology

Aditya Ghose

University of Wollongong

M Singh

North Carolina State University

Abstract

The media has been used to disseminate public information amid the Covid-19 pandemic. However, the Covid-19 news has triggered emotional responses in people that have impacted their mental well-being and led to news avoidance. To understand the emotional response to the Covid-19 news, we study user comments on the news published on Twitter by 37 media outlets in 11 countries from Jan 2020 to Dec 2022. We employ a deep-learning-based model to identify one of the 6 Ekman’s basic emotions, or the absence of emotional expression, in comments to the Covid-19 news, and an implementation of Latent Dirichlet Allocation (LDA) to identify 12 different topics in the news messages. Our analysis finds that while nearly half of the user comments show no significant emotions, negative emotions are more common. Anger is the most common emotion, particularly in the media and comments about political responses and governmental actions in the United States. Joy, on the other hand, is mainly linked to media outlets from the Philippines and news on vaccination. Over time, anger is consistently the most prevalent emotion, with fear being most prevalent at the start of the pandemic but decreasing and occasionally spiking with news of Covid-19 variants, cases, and deaths. Emotions also vary across media outlets, with Fox News having the highest level of disgust, the second-highest level of anger, and the lowest level of fear. Sadness is highest at Citizen TV, SABC, and Nation Africa, all three African media outlets. Also, fear is most evident in the comments to the news from The Times of India.

Keywords: COVID-19, News, Emotion, Twitter, media, Topic Modeling, deep learning

Suggested Citation

Oliveira, Francisco Braulio and Mougouei, Davoud and Haque, Amanul and Sichman, Jaime Simão and Dam, Hoa Khanh and Evans, Simon and Ghose, Aditya and Singh, M, Beyond Fear and Anger: A Global Analysis of Emotional Response to COVID-19 News on Twitter Using Deep Learning. Available at SSRN: https://ssrn.com/abstract=4383670 or http://dx.doi.org/10.2139/ssrn.4383670

Davoud Mougouei

Deakin University ( email )

75 Pigdons Road
Victoria, 3216
Australia

Amanul Haque

North Carolina State University ( email )

Jaime Simão Sichman

University of Sao Paulo ( email )

Hoa Khanh Dam

University of Wollongong ( email )

Simon Evans

University of Surrey - Department of Psychology ( email )

Aditya Ghose

University of Wollongong ( email )

M Singh

North Carolina State University

Do you have negative results from your research you’d like to share?

Paper statistics

Downloads
42
Abstract Views
154
PlumX Metrics