Source-linked AI summary
An analysis of Twitter messages in the 2011 Tohoku Earthquake
Son Doan, Bao-Khanh Ho Vo, Nigel Collier
TL;DR
This paper examines how Twitter reflected awareness and anxiety during the 2011 Tohoku earthquake, tsunami, and nuclear emergencies. Analyzing over 1.5 million English and Japanese tweets from Tokyo, it finds close correspondence with real-world events, similar patterns across languages, earlier native-language signals, and a rapid decline in anxiety.
Problem
The paper asks whether social media can reveal public opinion, awareness, and anxiety during natural disasters and support disaster preparedness and response.
Method
The authors analyze over 1.5 million English and Japanese Twitter messages from the Tokyo metropolitan area between 9th March and 31st May 2011.
Results
Twitter activity closely corresponded with earthquake events, while Japanese and English tweets showed correlated event patterns and Japanese tweets often peaked earlier.
Takeaways & Limitations
Twitter data can help track public mood during natural disasters and may support early warning through aggregated native-language messages.
Abstract
from arXiv · showhide
Social media such as Facebook and Twitter have proven to be a useful resource to understand public opinion towards real world events. In this paper, we investigate over 1.5 million Twitter messages (tweets) for the period 9th March 2011 to 31st May 2011 in order to track awareness and anxiety levels in the Tokyo metropolitan district to the 2011 Tohoku Earthquake and subsequent tsunami and nuclear emergencies. These three events were tracked using both English and Japanese tweets. Preliminary results indicated: 1) close correspondence between Twitter data and earthquake events, 2) strong correlation between English and Japanese tweets on the same events, 3) tweets in the native language play an important roles in early warning, 4) tweets showed how quickly Japanese people's anxiety returned to normal levels after the earthquake event. Several distinctions between English and Japanese tweets on earthquake events are also discussed. The results suggest that Twitter data can be used as a useful resource for tracking the public mood of populations affected by natural disasters as well as an early warning system.
1 Introduction
The paper examines whether Twitter can track public responses to the 2011 Tohoku earthquake and related disasters. Using English and Japanese tweets, it studies awareness and anxiety in Tokyo during the crisis.
- 1 Introduction: The work builds on prior uses of social media and web data for event detection, disaster alerts, and public-health surveillance.Examples include Twitter-based detection of influenza-like illness and disaster-related events.
- 1 Introduction: The Tohoku earthquake triggered a tsunami, extensive destruction, and nuclear accidents in Fukushima, creating a major crisis for Japan.The paper frames improved understanding of social responses as relevant to disaster preparedness and response.
- 1 Introduction: The study analyzes over 1.5 million Tokyo-area tweets to track social attitudes during the earthquake, tsunami, radiation emergency, and public anxiety.The observation period runs from 9 March through 31 May 2011.
- 1 Introduction: The analysis differentiates English and Japanese tweets to compare attitudes among local and foreign residents in metropolitan Tokyo.Tokyo experienced severe tremors, social anxiety, and mild radiation but no major loss of life.
2 Methods
The researchers collected geolocated Twitter messages from Tokyo over three months and separated the corpus by English and Japanese language.
- 2 Methods: The corpus contains 48,870 English tweets and 1,611,753 Japanese tweets collected from Tokyo between 9 March and 31 May 2011.Data were collected through the Twitter API using its geolocation feature.
b Earthquake events and relevant keywords
The analysis tracks earthquake and tsunami, radiation, and anxiety as indicators of public response to the disaster sequence. The events are examined across English and Japanese Twitter activity.
- b Earthquake events and relevant keywords: The study uses three indicators: earthquake and tsunami awareness, radiation from Fukushima Daiichi, and public anxiety.The first two indicators measure awareness, while the third examines anxiety in Tokyo.
- b Earthquake events and relevant keywords: The event sequence comprises the 11 March earthquake, the following tsunami, and the first Fukushima Daiichi reactor explosion on 12 March.The earthquake occurred at 05:46:23 UTC, followed by the tsunami minutes later; the first reactor explosion occurred at 06:36 UTC on 12 March.
- b Earthquake events and relevant keywords: Figure 1 reports tweet numbers by date separately for English and Japanese messages.The figure provides the temporal distribution used to examine activity around earthquake events.
- b Earthquake events and relevant keywords: Researchers manually constructed English and Japanese keyword lists for earthquake and tsunami, radiation, and public anxiety.They combined earthquake and tsunami because the corpus contained few English tweets about tsunami.
c Data analysis
Tweets are filtered with event-specific keywords and normalized by daily message volume to obtain relative event frequencies.
- c Data analysis: The analysis normalizes each event’s daily filtered-tweet count by the total number of tweets posted that day.This produces a relative frequency for each event per day.
- c Data analysis: Keyword filtering is used to identify tweets relevant to the earthquake, tsunami, radiation, and anxiety events.The event keywords are listed in Table 1.
3. Results and Discussions
Twitter activity closely tracked the earthquake, tsunami, radiation, and anxiety events in Tokyo, with Japanese and English messages often showing related patterns. Native-language tweets appeared earlier for several events, while anxiety declined toward stable levels after the initial crisis.
- Earthquake and tsunami event: The first Tokyo earthquake tweet appeared 1 minute and 25 seconds after the epicenter event, and Japanese tweets preceded English tweets by about 47 seconds.The corpus also records an early Tokyo tsunami retweet 12 minutes after the first tsunami report.
- Public response: Drill-down tweets captured practical effects and responses, including food shortages and recommendations to prepare emergency kits.The analysis also identified concerns about radiation in Tokyo tap water beginning 13 March.
- Earthquake and tsunami event: English and Japanese earthquake tweets correlated closely, indicating similar public concern during the earthquake events.The authors suggest these aggregated messages could support timely planning in future disasters.
- Radiation event: Japanese radiation tweets peaked on 12 and 15 March, while English tweets peaked one or two days later on 13 and 17 March.The authors interpret this timing difference as evidence that Japanese residents in Tokyo expressed radiation concern earlier than foreign residents.
- Anxiety event: Anxiety frequencies were highest on 11 March, declined over roughly two weeks, and then remained stable, with only a modest rise on 11 April.English tweets showed fewer messages but similar trends to Japanese tweets, especially from 20 April to 11 May.
4 Conclusions
The study finds strong correspondence between aggregated tweets and earthquake-related disasters, with native-language tweets contributing to early warning and anxiety tracking. The analysis remains open to broader evaluation and automated disaster-term detection.
- Aggregated tweets showed high correlations with earthquake, tsunami, and radiation events, supporting their use for disaster analysis.
- Native-language tweets played an important role in early warning through their volume and timeliness.
- The study identified potential for using Twitter to track public anxiety and needs among populations affected by disasters.
- Future work will extend the analysis to other earthquake aspects, apply publicly available evaluation metrics, and automate relevant-term discovery.