Twitter Trends following the Great East Japan Earthquake: A Quantitative Analysis of Earthquake-related Hashtags

Abstract

東日本大震災において Twitter によるコミュニケーションが注目を集めたが，本研究では震災後約 3 か月 にわたって収集した震災関連の 77 のハッシュタグのデータを元に，震災時の Twitter を用いたコミュニケー ションの実態を数値的に明らかにすることを目的とする．まず基本的な統計量から，多くのハッシュタグが震 災後徐々に利用が減少していく一方で，いくつかの話題は 4 月や 5 月がピークであることが分かった．また 自然言語処理の活用により，各ハッシュタグの話題の相違を分析した．さらにネットワーク分析の手法を用 いてハッシュタグ間の関係性の月ごとの変化を分析し，各ハッシュタグが組織化されていく傾向と中心的な 話題の変化を抽出した．これらの分析より災害時におけるハッシュタグの利用についての提言を行った． During the Great East Japan Earthquake, communication via Twitter attracted a great deal of attention. The aim of this paper is to quantify the actual level of Twitter-based communications related to the earthquake from data for 77 hashtags that was collected during a three-month period following the Great East Japan Earthquake. From some basic statistical analyses of the data, it is clear that, although there were a few earthquake-related peaks during April and May, the use of most hashtags gradually decreased following the earthquake. Utilizing NLP technology, differences in topic across the various hashtags were analyzed. Moreover, employing methods of social network analysis, the monthly changes in the relationships between the hashtags were analyzed and the organizational trends for each hashtag and the shifts in major topics were also extracted. Based on those analyses, some proposals are stated for the use of hashtags at the time of natural disasters.