Characterizing Behavior and Features of Participants and Observers during Disaster on Twitter

DOI: 10.5176/2251-1679_CGAT13.20

Authors: John Boaz Lee, Michael Yban ̃ez

Abstract:

The Philippines is a country that frequently expe- riences disasters, such as typhoons. During these events, many citizens spread information and communicate with each other through social media. This study aims to take advantage of that fact by analyzing the data from social media to get some insights on the situation.

Specifically, this paper studies the behavior and characteristics of Filipinos on Twitter during a disaster, and tries to see the differences between participants, or the direct victims of the disaster, and observers. The study used Latent Dirichlet Analysis to extract the different topics discussed during a disaster, and found out which topics participants are more likely to talk about. Results also show which topics are more likely to be retweeted, and which language participants in disaster use more often.