A comparative analysis of the COVID-19 Infodemic in English and Chinese: insights from social media textual data

Jia Luo, Daiyun Peng, Lei Shi, Didier El Baz, Xinran Liu

Abstract

The COVID-19 infodemic, characterized by the rapid spread of misinformation and unverified claims related to the pandemic, presents a significant challenge. This paper presents a comparative analysis of the COVID-19 infodemic in the English and Chinese languages, utilizing textual data extracted from social media platforms. To ensure a balanced representation, two infodemic datasets were created by augmenting previously collected social media textual data. Through word frequency analysis, the 30 most frequently occurring infodemic words are identified, shedding light on prevalent discussions surrounding the infodemic. Moreover, topic clustering analysis uncovers thematic structures and provides a deeper understanding of primary topics within each language context. Additionally, sentiment analysis enables comprehension of the emotional tone associated with COVID-19 information on social media platforms in English and Chinese. This research contributes to a better understanding of the COVID-19 infodemic phenomenon and can guide the development of strategies to combat misinformation during public health crises across different languages.

Introduction

During the COVID-19 pandemic, the world faced a secondary crisis: the "infodemic." This surge of both accurate and false information spread as rapidly as the virus itself, complicating public health responses. While many studies have attempted to detect fake news, they often focus exclusively on English-language data and rely on binary "true vs. false" classifications. This research addresses these gaps by performing a comparative analysis between the world’s two most common internet languages: English and Chinese. Furthermore, the authors introduce an "uncertain" category, acknowledging that during a fast-moving crisis, the veracity of many claims is not immediately clear.

Methodology

The researchers employed a multi-pronged Natural Language Processing (NLP) pipeline to analyze social media text across two primary datasets:

  • English Dataset: Sourced from a 2021 study by Patwa et al., consisting of 2,490 records (830 each of True, False, and Uncertain) after healthcare worker verification.
  • Chinese Dataset: Sourced from a 2021 study by Luo et al., containing 1,055 records (335 True, 276 False, and 409 Uncertain) gathered from platforms like Weibo and WeChat.

To ensure data quality, healthcare workers manually re-classified thousands of records. The analysis included Word Frequency Analysis using the Weiciyun tool, Topic Clustering via Latent Dirichlet Allocation (LDA) to uncover hidden thematic structures, and Sentiment Analysis using Monkeylearn (English) and ROST_CM6 (Chinese).

Comparative Analysis of Discourse

The study revealed distinct cultural and political differences in how the infodemic manifested in each language. English discourse was found to be more politically polarized, with President Donald Trump being the most frequently mentioned individual. In contrast, Chinese records centered on medical authority, with respiratory expert Zhong Nanshan appearing most prominently.

Figure 1: Word clouds illustrating the most frequent terms in English and Chinese COVID-19 social media records.

Cultural nuances in prevention were also evident; the term "mask" was significantly more prominent in Chinese word clouds, reflecting early proactive adoption compared to the more debated stance in Western contexts. Furthermore, English records tended to focus on real-time temporal updates (e.g., "today," "yesterday"), while Chinese records were more geographically specific, frequently mentioning cities like Wuhan, Beijing, and Shanghai.

Key Results and Sentiment Analysis

  • Prevalent Negativity: Both languages were dominated by negative sentiment. English records were 59.96% negative, while Chinese records were 50.71% negative.
  • Sentiment Polarization: English speakers showed more extreme emotional variance, possessing both higher negative sentiment and a higher proportion of positive sentiment (31.12%) compared to Chinese (25.69%). Chinese discourse was characterized as more "neutral and conservative."
  • Misinformation and Emotion: In both languages, records labeled as "false" exhibited the highest proportion of negative sentiment, confirming a strong link between misinformation and the generation of fear or anger.
  • Thematic Diversity: The English "false" group contained a significantly higher number of clustered topics than the Chinese group, suggesting that the English-speaking web hosted a wider variety of distinct false narratives and conspiracy theories.

Conclusion

This research highlights that while the COVID-19 infodemic is a global phenomenon, its content and emotional tone are deeply influenced by linguistic and cultural contexts. While the study is limited by source bias (relying on fact-checking channels) and a lack of propagation data, it provides a foundation for future work. Next steps involve incorporating social engagement metrics—such as reposts and replies—to model how these rumors travel through populations over time.

Summary updated

Abstract

The COVID-19 infodemic, characterized by the rapid spread of misinformation and unverified claims related to the pandemic, presents a significant challenge. This paper presents a comparative analysis of the COVID-19 infodemic in the English and Chinese languages, utilizing textual data extracted from social media platforms. To ensure a balanced representation, two infodemic datasets were created by augmenting previously collected social media textual data. Through word frequency analysis, the 30 most frequently occurring infodemic words are identified, shedding light on prevalent discussions surrounding the infodemic. Moreover, topic clustering analysis uncovers thematic structures and provides a deeper understanding of primary topics within each language context. Additionally, sentiment analysis enables comprehension of the emotional tone associated with COVID-19 information on social media platforms in English and Chinese. This research contributes to a better understanding of the COVID-19 infodemic phenomenon and can guide the development of strategies to combat misinformation during public health crises across different languages.
BibTeX
@inproceedings{luo2023comparative,
  author = {Jia Luo and Daiyun Peng and Lei Shi and Didier El Baz and Xinran Liu},
  title = {{A comparative analysis of the COVID-19 Infodemic in English and Chinese: insights from social media textual data}},
  booktitle = {Frontiers in Public Health},
  year = {2023},
  doi = {10.3389/fpubh.2023.1281259},
}
Figure