Once a disaster occurs, people discuss various topics in social media such as electronic bulletin boards, SNSs and video services, and their decision-making tends to be affected by discussions in social media. Under the circumstance, a mechanism to detect topics in social media has become important. This paper targets the East Japan Great Earthquake, and proposes a time series topic transition discovering method in social media. Our proposed method adopts directed graphs to show topic structures in social media, and then form clusters using modularity measure which expresses the quality of a division of a network into modules or communities. The method computes topic transition using the Matthews correlation coefficient which is a measure of the quality of two binary classifications, and analyzes them over time. An experimental result using actual social media data about the East Japan Great Earthquake is shown as well.
