Abstract
The objective of this study is to clarify the relationship between social media information (i.e., Twitter) and returning home behavior in disaster. In order to analyze Twitter's textual data, which is tweeted in the Great East Japan Earthquake on 11 March, 2011, this study applies text-mining approach. The results of text-mining confirm the trend of collocation's network among terms after the Earthquake happened, and the distribution of term co-occurrence frequency. On the other hand, using the returning home data sets on 11 March, 2011 collected at Tokyo and Kanagawa, this study calculate the differences of estimated population between one point and the former point. Moreover, using the two data sets, this study analyzed the relationship between Social Media and returning home behavior using a canonical correlation analysis. Term co-occurrence frequency in Social Media and behavior of returning home is applied to two sets of variables in canonical analysis, and the estimation results demonstrated that these two are significantly interdependent.
