Emotion analysis and detection during COVID-19

TL;DR

This paper analyzes emotional expressions during COVID-19 using Twitter data, revealing emotional impacts over time and exploring how large language models can predict emotions with semi-supervised learning.

Contribution

It introduces CovidEmo, a dataset of 3,000 labeled tweets, and evaluates cross-domain generalization of language models, proposing semi-supervised learning to improve emotion prediction.

Findings

Large language models show some cross-domain transfer but have notable gaps.

Semi-supervised learning with unlabeled data improves emotion prediction accuracy.

Emotional impact of COVID-19 varies over time and influences social narratives.

Abstract

Crises such as natural disasters, global pandemics, and social unrest continuously threaten our world and emotionally affect millions of people worldwide in distinct ways. Understanding emotions that people express during large-scale crises helps inform policy makers and first responders about the emotional states of the population as well as provide emotional support to those who need such support. We present CovidEmo, ~3K English tweets labeled with emotions and temporally distributed across 18 months. Our analyses reveal the emotional toll caused by COVID-19, and changes of the social narrative and associated emotions over time. Motivated by the time-sensitive nature of crises and the cost of large-scale annotation efforts, we examine how well large pre-trained language models generalize across domains and timeline in the task of perceived emotion prediction in the context of COVID-19. Our analyses suggest that cross-domain information transfers occur, yet there are still significant gaps. We propose semi-supervised learning as a way to bridge this gap, obtaining significantly better performance using unlabeled data from the target domain.