Abstract
During COVID-19, misinformation on social media affects the adoption of appropriate prevention behaviors. It is urgent to suppress the misinformation to prevent negative public health consequences. Although an array of studies has proposed misinformation suppression strategies, few have investigated the role of predominant credible information during crises. None has examined its effect quantitatively using longitudinal social media data. Therefore, this research investigates the temporal correlations between credible information and misinformation, and whether predominant credible information can suppress misinformation for two prevention measures (i.e. topics), i.e. wearing masks and social distancing using tweets collected from February 15 to June 30, 2020. We trained Support Vector Machine classifiers to retrieve relevant tweets and classify tweets containing credible information and misinformation for each topic. Based on cross-correlation analyses of credible and misinformation time series for both topics, we find that the previously predominant credible information can lead to the decrease of misinformation (i.e. suppression) with a time lag. The research findings provide empirical evidence for suppressing misinformation with credible information in complex online environments and suggest practical strategies for future information management during crises and emergencies.
Introduction
During the COVID-19 pandemic, misinformation emerged as a significant public health threat, hindering the adoption of life-saving behaviors such as mask-wearing and social distancing. While social media platforms often rely on reactive strategies like fact-checking or banning accounts, this study investigates a proactive alternative: Information Suppression. The research explores whether flooding the digital environment with credible information from authoritative sources can naturally reduce the spread of misinformation over time.
Methodology
The researchers developed a multi-stage pipeline to transform raw social media data into statistical evidence. This process involved data retrieval via the Twitter Streaming API and Hydrator, followed by a rigorous preprocessing and classification phase.
The pipeline utilized Support Vector Machine (SVM) classifiers with TF-IDF vectorization to categorize tweets by topic and information quality (credible vs. misinformation). To analyze the temporal relationship between these categories, the study employed Cross-Correlation Function (CCF) analysis and ARIMA models for pre-whitening, ensuring that observed correlations were not merely artifacts of shared external trends.
Data and Classifier Performance
The study analyzed a massive corpus of 28,573,952 English tweets collected between February 15 and June 30, 2020. From this dataset, subsets were extracted for "Wearing Masks" (667,761 tweets) and "Social Distancing" (101,113 tweets).
To ensure the reliability of the findings, the SVM classifiers were trained on manually tagged datasets. The performance metrics indicate high reliability in distinguishing between credible information and misinformation across both topics.
As shown in the table, the classifiers achieved accuracy rates ranging from approximately 88% to 91% for the final information classification stage, with high precision and recall scores. This quantitative foundation allowed for robust longitudinal analysis.
Key Insights and Results
The analysis revealed several critical findings regarding the dynamics of online information:
- The Battlefield Effect: On a daily basis, credible information and misinformation often show a positive correlation, suggesting that both sides ramp up messaging simultaneously in response to major events.
- Lagged Suppression: The core discovery is the "lagged victory" of credible information. A high volume or proportion of credible information today leads to a significant decrease in misinformation with a typical 1-day lag.
- Robustness Across Topics: While the "wearing masks" topic was more volatile and high-volume, the suppression effect remained consistent for "social distancing," indicating a robust phenomenon across different public health measures.
Conclusion
This research provides empirical evidence that predominant credible information can act as a proactive suppressant for misinformation during crises. Public health agencies are encouraged to maintain a consistent presence of authoritative facts to create an environment where misinformation struggles to take root. Future work may extend these models to other languages, platforms, and advanced AI architectures like Transformers to further refine classification and suppression strategies.
Summary updated