TEXT MINING ON REAL TIME TWITTER DATA FOR DISASTER RESPONSE

Social media such as micro blogging services have a significant impact on the day-to-day lives of people. These services are currently being used by government agencies to interact and communicate information to general public. They also bring an effective collaboration of all stakeholders for dissemination of information during an emergency. Social media is capable of providing spontaneous information during emergency/disaster situations unlike news media, therefore, particularly micro blogging services, have the potential to be adopted as an additional tool for emergency services. In the present work the authors by mining real time data from twitter TM tried to predict the impending damage in the following days during flood scenario. The users of twitter provide important information such as warnings, location of an event, first hand experiences. Such information is collected, preprocessed, geo located and filtered. From the collected information, geo-coded data is prioritized to that of text data. Then the data is analyzed to find the course of the disaster through regression analysis. Later, disaster curve is extrapolated for prediction of damage susceptible locations in the following days. The results are validated by analyzing the past events. In this study, 2015 Chennai flood data is used to validate the results. The study has the potential to facilitate disaster managers for better response operations during emergencies