AN ENHANCED MODEL FOR DISASTER MANAGEMENT USING SOCIAL NETWORK AND MACHINE LEARNING TECHNIQUES

Abstract

Millions of posts in the form of text, images and videos are being posted everyday on social media and platforms like Twitter, Facebook plays a very critical role in spreading and sharing of the information across the world. Due to this, social media can have an active role to play in the case of natural disasters. During and after natural disasters, a large number of users posts information regarding the disaster and their situation on the social platforms. These platforms are also actively used by the government & other agencies to share critical information. According to a survey, social media ranks fourth most popular source for accessing emergency information. Although due to the lack of right information among the huge volume of the incoming tweets during any disaster, the resources are not properly mapped which leads to further loss of life & property which can be prevented to a certain extent. Large volume of the data present on social media can be leveraged to access the disaster situations and prepare a management plan accordingly. In this work we study how the data present on social media can be helpful to manage the disaster situation. Efficient analysis of the data can help in preparing a plan for proper information propagation. Most critical part is to mine the right data in a quick time from the huge volume of data posted by the users. So in our first task we plan to classify the tweets obtained during the disaster situation and separate the tweets which are related to the ongoing situation. Doing this reduces the number of tweets to be analyzed further to get the important information. After the identification of the disaster related tweets, it is important to understand the need and emotions of the user, thus we plan to do sentiment analysis over the tweets and gather the information about the needs of the users which can be used to provide assistance for their recovery.