Language:
Search
 

Ministry of Economy, Trade and Industry
Japan

The Problem

Telecommunication infrastructure was seriously damaged or washed away in the devastated Tohoku region, it exacerbated official attempts to determine the actual situation and clarify the necessary assistance. (Having difficulties in understanding what relief supply was needed by the victims in evacuation centers.) In addition, telephone lines were congested extensively, not only in the devastated Tohoku region, but also in the Tokyo metropolitan area because so many calls were made by large number of commuters unable to get home caused by traffic paralysis.
In such situation, the Internet served as an important communication tool. Especially, SNS was used as the primary mode of communication, from requesting medical aid, assistance, seeking information about missing people, sending encouragement and also reporting damage and transportation infrastructure statuses. .
In addition, a widening circle of cooperation has spread among Internet users in response to the pleas for help on SNS.

Solution and Key Benefits

 What is the initiative about? (the solution)
Text Analysis in the emergency situation

METI E-Government team was implementing the text analysis trial project at the time. Using the key word "lack", we retrieved about 1000 tweets by Google each day from March 12-15 and about 1500 tweets by a visualization engine from March 16-26. A month after the earthquake, the number of tweets we retrieved increased up to about 200 000. The result from this analysis was prominent data to determine relief supply to the devastated region. In addition, we extracted public opinion from social media to capture large-scale trends of public demands to the government. We took the following three measures.

1. In response to acute shortages of dry cell batteries, we used text analysis to assess exact information about the battery insufficiency, such as size and type. In this analysis, we retrieved Twitter messages containing a topic key word “battery” for the text data.
2. Based on the Twitter data, we tracked the need trend for insufficient things and determine the necessary assistance.
3. Using Twitter messages containing “lack” posted after the earthquake, collected by querying the Twitter API, we prepared "Daily Ranking of the Need Trend for Insufficient Things"

In addition, we took the following two measures, since the measures we needed to take changed by time.
1. Analysis of the reliability of the government
2. Analysis of rumors

Actors and Stakeholders

 Who proposed the solution, who implemented it and who were the stakeholders?
METI E-Government team implemented the initiative, from proposal, implementation, to analysis. Using tools in place for another project which was still in the trial and error stage, we tried various kinds of analysis. Because we from usually used sns, it was possible to plan the combination between text analysis and sns.

(a) Strategies

 Describe how and when the initiative was implemented by answering these questions
 a.      What were the strategies used to implement the initiative? In no more than 500 words, provide a summary of the main objectives and strategies of the initiative, how they were established and by whom.
Our main objective was to determine the actual situation of the devastated region in a situation where telecommunication infrastructure had been seriously damaged or washed away. At the same time, extract public opinion so that the government could take measures properly.
Our strategy was to examine the effectiveness of SNS and text analysis to capture the whole context of emergency situation, by so doing, enhance the crisis management system in Japan.
（How they were established and by whom）
Not being able to get real-time information about the devastated Tohoku area since telecommunication infrastructure had been seriously damaged due to the Tsunami, we realized, “a visualization engine”, a tool used in another project still in experimental stage, to get information. Although we knew that we could use this to analyze SNS messages, we had never encountered such an occasion to utilize the function in our project before the earthquake. After the earthquake, however, the situation surrounding SNS changed drastically, in turn, the social fabric of the nation quickly revolved around the use of Twitter as the primary mode of communication. For example, while Twitter was used predominantly to talk about entertainment (about 60% of the topics were entertainment) before the earthquake, it quickly morphed into something entirely different on the day of the disaster, where 72% of the topics were related to the Earthquake, and another 8% were on transportation. In some ways Twitter became the virtual bulletin board for exchanging valuable information, disseminating it to the public, and utilizing the social networks to “spread to the word quickly and effectively. For March 11 alone, 33 million tweets were reported in Japan, almost double the average daily usage. In response to such situation after the Earthquake, we analyzed the data from SNS messages seeking to understand what was needed after the earthquake and analyzed the reliability of the government.

(b) Implementation

 b.      What were the key development and implementation steps and the chronology? No more than 500 words
（March 11）
The Great East Japan Earthquake
（March 14）
In response to acute shortage of dry cell batteries, we used text analysis to assess exact information about the battery insufficiency, such as size and type. In this analysis, we retrieved Twitter messages containing a topic key word “battery” for the text data.
Retrieving Twitter messages containing a topic key word “assistance” and “need” for the text data. Based on the result, we analyzed the categories of insufficient things and assistance required urgently in the devastated area.
Retrieving Twitter messages containing “lack” posted after the earthquake, collected by querying the Twitter API, we prepared "Daily Ranking of the Need Trend for Insufficient Things"
（April 11）
“Text Analysis Guidebook”
METI (IT Project Office) prepared for a guidebook for encouraging the efficient use of text analysis.
（May 8）
METI (IT Project Office) practiced analysis for the reliability of the government.
METI (IT Project Office) practiced analysis for rumors and bad reputation of products made in Tohoku region in response to nuclear accident. Food and products produced around Tohoku area might have been contaminated with radioactive material; it caused people to avoid not only food and products made in Tohoku region but also visiting the region.

(c) Overcoming Obstacles

 c.      What were the main obstacles encountered? How were they overcome? No more than 500 words
The main obstacles encountered were the respondent’s and response bias depending on time because the number of Twitter messages you can get at one time is limited to 1500. At the same time, the data capacity was limited because we used cloud service. To solve response bias, we tried to get Twitter message many times in a day and used the function to delete multiple re-tweet. We also limit the analysis target and made a back up copy because the data capability was limited.

(d) Use of Resources

 d.      What resources were used for the initiative and what were its key benefits? In no more than 500 words, specify what were the financial, technical and human resources’ costs associated with this initiative. Describe how resources were mobilized
Our resources were 5 staff members from METI e-government team. We used text analysis tool utilizing cloud computing, which costs about 1.4 million Japanese yen per half year. The fundamental benefit is that it increased the productivity of the staff. Considering it might have taken several weeks to analyze the urgent demands of the devastated area using traditional methods, text analysis of SNS, which takes only 30 minutes to get a result, was highly efficient to determine the actual situation of the devastated region in minutes.

Sustainability and Transferability

  Is the initiative sustainable and transferable?
The initiative is sustainable and transferable since the tool is not expensive however experience and know-how are required.
Considering it is crucial to accumulate know-how, we prepared a guidebook based on this experience to introduce the tool to other public agencies. We are now reviewing the analysis methods, including usage examples in the private sector. In this way, we are taking measures in which other organizations can implement the text analysis using the tool we used this time. In addition, we provide information actively about our project and activities through lectures and articles.

Lessons Learned

 What are the impact of your initiative and the lessons learned?
In response to this Earthquake, we could the capture real-time situation of the devastated area within 30 minutes under the conditions where the telecommunication infrastructure was seriously destroyed. It suggests that text analysis could be a very useful and efficient tool in crisis management. Learning from this experience, we are considering implementing profile text analysis including analysis by gender, age and area. In the future, the government is expected to be able to respond appropriately in emergency situations using text analysis.
Moreover, the result revealed that the text analysis increased productivity of the staff members our team and improved the sophistication of public service. At the same time, one of the greatest impacts was that the momentum to use text analysis was fostered in response to our activity.

Contact Information

Institution Name:   Ministry of Economy, Trade and Industry
Institution Type:   Government Agency  
Contact Person:   Kenji Hiramoto
Title:   Exuecutive adviser for CIO  
Telephone/ Fax:   +81-3-3501-2964/+81-3-3580-6403
Institution's / Project's Website:   http://openlabs.go.jp/english/
E-mail:   hiramoto-kenji@meti.go.jp  
Address:   1-3-1 Kasumigaseki Chiyoda-ku
Postal Code:   1008901
City:   Chiyoda-ku
State/Province:   Tokyo
Country:   Japan

          Go Back
Print friendly Page