Lies Kill, Facts Save: Detecting COVID-19 Misinformation in Twitter

Area of Science:

• Computational Social Science
• Public Health Informatics
• Artificial Intelligence

Background:

• Online social networks (ONSs) like Twitter are vital for information dissemination but also facilitate the rapid spread of misinformation.
• The coronavirus disease 2019 (COVID-19) pandemic has been accompanied by a significant 'infodemic', highlighting the urgent need for scientific fact-checking.
• Detecting and mitigating the spread of false information on ONSs is critical for public health during global crises.

Purpose of the Study:

• To analyze the credibility of information shared on Twitter concerning the COVID-19 pandemic.
• To propose and evaluate an ensemble-learning-based framework for verifying the credibility of a large volume of COVID-19 related tweets.
• To classify tweets into 'credible' or 'non-credible' categories based on tweet- and user-level features.

Main Methods:

• Collected and labeled a large dataset of tweets related to COVID-19.
• Developed an ensemble-learning framework incorporating tweet-level and user-level features for credibility assessment.
• Conducted multiple experiments to validate the performance of the proposed framework.

Main Results:

• The proposed framework demonstrated high accuracy in classifying tweets as either credible or non-credible.
• The ensemble-learning approach effectively leveraged various features to discern information veracity.
• The study successfully identified patterns distinguishing credible from non-credible COVID-19 information on Twitter.

Conclusions:

• The developed framework is effective for detecting misinformation regarding the COVID-19 pandemic on Twitter.
• Ensemble learning provides a robust method for large-scale credibility analysis of social media content.
• Accurate detection of COVID-19 misinformation is essential for mitigating public health risks associated with the infodemic.