Estimating the Duration of Public Concern After the Fukushima Dai-ichi Nuclear Power Station Accident From the Occurrence of Radiation Exposure-Related Terms on Twitter: A Retrospective Data Analysis

How Italy Tweeted about COVID-19: Detecting Reactions to the Pandemic from Social Media

The COVID-19 pandemic required communities throughout the world to deal with unknown threats. Using Twitter data, this study aimed to detect reactions to the outbreak in Italy and to evaluate the relationship between measures derived from social media (SM) with both national epidemiological data and reports on the violations of the restrictions. The dynamics of time-series about tweets counts, emotions expressed, and themes discussed were evaluated using Italian posts regarding COVID-19 from 25 February to 4 May 2020. Considering 4,988,255 tweets, results highlight that emotions changed significantly over time with anger, disgust, fear, and sadness showing a downward trend, while joy, trust, anticipation, and surprise increased. The trend of emotions correlated significantly with national variation in confirmed cases and reports on the violations of restrictive measures. The study highlights the potential of using SM to assess emotional and behavioural reactions, delineating their possible contribution to the establishment of a decision management system during emergencies.

Relationships Among Tweets Related to Radiation: Visualization Using Co-Occurring Networks

BACKGROUND

After the Fukushima Daiichi nuclear accident on March 11, 2011, interest in, and fear of, radiation increased among citizens. When such accidents occur, appropriate risk communication must provided by the government. It is therefore necessary to understand the fears of citizens in the days after such accidents.

OBJECTIVE

This study aimed to identify the progression of people's concerns, specifically fear, from a study of radiation-related tweets in the days after the Fukushima Daiichi nuclear accident.

METHODS

From approximately 1.5 million tweets in Japanese including any of the phrases "radiation" (), "radioactivity" (), and "radioactive substance" () sent March 11-17, 2011, we extracted tweets that expressed fear. We then performed a morphological analysis on the extracted tweets. Citizens' fears were visualized by creating co-occurrence networks using co-occurrence degrees showing relationship strength. Moreover, we calculated the Jaccard coefficient, which is one of the co-occurrence indices for expressing the strength of the relationship between morphemes when creating networks.

RESULTS

From the visualization of the co-occurrence networks, we found high citizen interest in "nuclear power plant" on March 11 and 12, "health" on March 12 and 13, "medium" on March 13 and 14, and "economy" on March 15. On March 16 and 17, citizens' interest changed to "lack of goods in the afflicted area." In each co-occurrence network, trending topics, citizens' fears, and opinions to the government were extracted.

CONCLUSIONS

This study used Twitter to understand changes in the concerns of Japanese citizens during the week after the Fukushima Daiichi nuclear accident, with a focus specifically on citizens' fears. We found that immediately after the accident, the interest in the accident itself was high, and then interest shifted to concerns affecting life, such as health and economy, as the week progressed. Clarifying citizens' fears and the dissemination of information through mass media and social media can add to improved risk communication in the future.