Use of Social Media/Microblogging Platform along with GIS for Mapping Disaster

摘要

Twitter is a vast store of textual data that can be helpful in determining disaster related crisis. In this research the aim is to implement a system that can acquire tweets from Twitter using the keyword `earthquake' and determine the disaster location soon after the disaster. For prefiltering the tweets, a statistical approach has been developed. For determining out of context tweets a fuzzy matching approach has been used. To further categorize those tweets as relevant and irrelevant, Multinomial Na?ve Bayes classifier has been used. Relevant tweets were further analyzed to determine probable location of the disaster using a clustering-based approach. After obtaining the probable location of the disaster; it is then visualized on map.