[Paper Review] Towards Characterizing COVID-19 Awareness on Twitter

Muhammad Saad, Muhammad Hassan+1 more|arXiv (Cornell University)|May 17, 2020

Misinformation and Its Impacts10 references7 citations

TL;DR

This study analyzes Twitter data from 20 top COVID-19-affected countries (Dec 2019–Apr 2020) to characterize public awareness and its relationship with pandemic spread. Using trend tracking, tweet volume analysis, topic modeling, and sentiment analysis, it finds that countries with lower infection rates generated more timely and positive discussions about preventive measures like social distancing and quarantine, suggesting social media can amplify public health awareness and potentially influence pandemic outcomes.

ABSTRACT

The coronavirus (COVID-19) pandemic has significantly altered our lifestyles as we resort to minimize the spread through preventive measures such as social distancing and quarantine. An increasingly worrying aspect is the gap between the exponential disease spread and the delay in adopting preventive measures. This gap is attributed to the lack of awareness about the disease and its preventive measures. Nowadays, social media platforms (ie., Twitter) are frequently used to create awareness about major events, including COVID-19. In this paper, we use Twitter to characterize public awareness regarding COVID-19 by analyzing the information flow in the most affected countries. Towards that, we collect more than 46K trends and 622 Million tweets from the top twenty most affected countries to examine 1) the temporal evolution of COVID-19 related trends, 2) the volume of tweets and recurring topics in those trends, and 3) the user sentiment towards preventive measures. Our results show that countries with a lower pandemic spread generated a higher volume of trends and tweets to expedite the information flow and contribute to public awareness. We also observed that in those countries, the COVID-19 related trends were generated before the sharp increase in the number of cases, indicating a preemptive attempt to notify users about the potential threat. Finally, we noticed that in countries with a lower spread, users had a positive sentiment towards COVID-19 preventive measures. Our measurements and analysis show that effective social media usage can influence public behavior, which can be leveraged to better combat future pandemics.

Motivation & Objective

• To investigate how public awareness of COVID-19 evolved on Twitter across different countries during the early pandemic.
• To examine whether the volume and timing of Twitter trends and tweets correlate with national pandemic spread rates.
• To analyze user sentiment toward preventive measures such as social distancing, quarantine, and lockdown policies.
• To identify recurring topics in tweets to understand public discourse on prevention and risk mitigation.
• To assess whether social media activity can serve as an early indicator of effective public health preparedness.

Proposed method

• Collected over 48,000 Twitter trends and 622 million tweets from the top 20 countries most affected by COVID-19 as of April 19, 2020.
• Deployed a distributed crawler system using Trendogate to collect trends and a scheduler to distribute tweet collection tasks to 85 workers.
• Applied natural language processing (NLP) techniques including bigram analysis and topic modeling to extract recurring themes from tweets.
• Conducted sentiment analysis using a pre-trained model to classify sentiment toward key preventive measures (social distancing, quarantine, lockdown).
• Correlated temporal patterns of trends and tweet volumes with official national COVID-19 case timelines to assess preparedness.
• Performed a comparative case study across six countries (USA, Italy, Spain, Sweden, Austria, Belgium) to highlight cross-national differences in awareness and sentiment.

Experimental results

Research questions

• RQ1 Do variations in Twitter engagement (trends and tweet volume) reflect differences in national preparedness for the COVID-19 pandemic?
• RQ2 Is there a temporal correlation between the emergence of COVID-19-related trends and the subsequent rise in confirmed cases?
• RQ3 How do user sentiments toward preventive measures like social distancing and lockdown vary across countries with differing pandemic outcomes?
• RQ4 What are the dominant topics discussed in tweets related to COVID-19 in countries with lower versus higher infection rates?
• RQ5 Can social media activity serve as a proxy for public awareness and potentially influence pandemic spread dynamics?

Key findings

• Countries with lower pandemic spread generated significantly higher volumes of COVID-19-related trends and tweets, suggesting proactive public awareness on social media.
• In countries with lower spread, COVID-19-related trends emerged before the sharp rise in cases, indicating preemptive information dissemination.
• Users in low-spread countries expressed more positive sentiment toward preventive measures, with sentiment peaks around 0.9 for social distancing in Austria and Belgium.
• Sentiment toward lockdown policies varied significantly: Italy showed a negative sentiment peak around -0.1, while Austria and Belgium showed positive peaks around 0.9.
• Topic modeling revealed that 'social distancing' was the most dominant term in Sweden and Austria, while 'self-quarantined' was prominent in Spain.
• The overall sentiment for social distancing and quarantine was uniformly positive across countries, but sentiment toward lockdown showed notable divergence, possibly due to societal or policy-related factors.