Public Discourse on Environmental Pollution and Health in Korea: Tweets Following the Fukushima Nuclear Accident

Seung‐Hoi Kim, Yu-i Ha, Meeyoung Cha, Jiyon Lee, Byoung-Jik Kim, Dong-Myung Lee

Abstract

Public discourse on environmental and health issues has risen on social media. Upon an environmental crisis, various chatters such as breaking news, misinformation, and rumor could aggravate social confusion and proliferate negative public sentiment. In an effort to study public sentiments on environmental issues in South Korea, we analyzed 158,964 tweets generated over a 4-year period following the Fukushima accident in 2011, the largest release of radioactivity to environment in recent history. This event led to a significant increase in public’s interest on environmental and nuclear issues in Korea. We employed Bayesian network and recursive partitioning to observe the classification regression tree structure of major topics. Topics on health and environment were interlinked closely and represented both apprehension and concern about health threats and pollution. Our methodology helps analyze large online discourse efficiently and offers insight to crisis response organizations.

Introduction

When environmental disasters strike, social media becomes a vital source of real-time information but also a breeding ground for misinformation and social panic. Following the 2011 Fukushima Daiichi nuclear disaster, South Korea—Japan’s closest neighbor—experienced a surge in public anxiety regarding radiation. This study addresses the challenge of making sense of massive volumes of "online chatter" to help crisis response organizations understand how public interest shifted from immediate disaster news to long-term concerns about food safety and chronic illness.

Methodology

The researchers analyzed a custom dataset of 158,964 Korean-language tweets spanning from January 1, 2011, to December 14, 2014. The analytical pipeline involved gathering tweets using keywords like "Radioactive" and "Nuclear Power," followed by manual coding to identify 43 key terms. These terms were categorized into four themes: Location, Health, Environment, and Nuclear Power.

Table 1: Categorization of keywords into four major topics and their relative frequency in the dataset.

Beyond simple keyword counts, the study utilized Bayesian networks and recursive partitioning to map dependencies between terms. This allowed researchers to observe the "logic" of public concern—for example, how a tweet about "imports" was probabilistically linked to "fishery products" and subsequently to health outcomes.

Results and Discussion

The study revealed a significant shift in public discourse. The average number of daily tweets regarding nuclear issues jumped from 10 per day before the accident to over 100 per day after. "Nuclear Power" was the most discussed category (37.2%), followed by "Environment" (23.3%) and "Health" (20.9%).

The "Food-to-Health" Pipeline

A key finding was the tight coupling between environmental pollution and specific health fears. The Bayesian analysis revealed a clear mental map: concerns followed a specific path from environmental causes to health effects. For instance, tweets about "imports" were probabilistically linked to "fishery products" and subsequently to "cancer."

Figure 1: Recursive partitioning tree showing the dependencies between 'Pollution', 'Import', and 'Fishery Product' topics.

Communication Gaps

The research highlighted a significant "communication gap" between the public and government organizations. While citizens were discussing thyroid cancer and food safety, government agencies often used Twitter for unrelated promotional activities. This "tone-deaf" approach represents a missed opportunity for effective risk communication and rumor management during a crisis.

Conclusion

This longitudinal study demonstrates that health and environment topics are statistically inseparable in the public mind during a nuclear crisis. While the study provides a robust framework for analyzing large-scale discourse, future work could benefit from incorporating user demographics and deeper sentiment analysis to provide more targeted policy insights. Crisis response organizations should focus on addressing the specific "keyword paths"—such as the link between food imports and health risks—that drive public anxiety.

Reference: Kim, S. H., Ha, Y. I., Cha, M., Lee, J., Kim, B. J., & Lee, D. M. (2016). Public Discourse on Environmental Pollution and Health in Korea: Tweets Following the Fukushima Nuclear Accident. Technical Report WS-16-20, AAAI Conference on Web and Social Media.

Summary updated

Abstract

Public discourse on environmental and health issues has risenon social media. Upon an environmental crisis, various chatterssuch as breaking news, misinformation, and rumor couldaggravate social confusion and proliferate negative publicsentiment. In an effort to study public sentiments on environmentalissues in South Korea, we analyzed 158,964 tweetsgenerated over a 4-year period following the Fukushima accidentin 2011, the largest release of radioactivity to environmentin recent history. This event led to a significant increasein public’s interest on environmental and nuclear issues inKorea. We employed Bayesian network and recursive partitioningto observe the classification regression tree structureof major topics. Topics on health and environment were interlinkedclosely and represented both apprehension and concernabout health threats and pollution. Our methodologyhelps analyze large online discourse efficiently and offers insightto crisis response organizations.
BibTeX
@inproceedings{kim2021public,
  author = {Seung‐Hoi Kim and Yu-i Ha and Meeyoung Cha and Jiyon Lee and Byoung-Jik Kim and Dong-Myung Lee},
  title = {{Public Discourse on Environmental Pollution and Health in Korea: Tweets Following the Fukushima Nuclear Accident}},
  booktitle = {AAAI Conference on Artificial Intelligence},
  year = {2021},
  doi = {10.1609/icwsm.v10i2.14854},
}
Figure