Mining conversations of geographically changing users

摘要: In recent disaster events, social media has proven to be an effective communication tool for affected people. The corpus of generated messages contains valuable information about the situation, needs, and locations victims. We propose approach extract significant aspects user discussions better inform responders enable appropriate response. methodology combines location based division users together with standard text mining (term frequency inverse document frequency) identify important topics conversation in a dynamic geographic network. further suggest that both movement patterns change during disaster, which requires identification new trends. When applied area suffered this can provide 'sensemaking' through insights into where people are located, they going what communicate when moving.