Social media text mining for decision support in natural disaster management in Sri Lanka

Abstract

With the popularity of internet and smart devices, social media is very popular today among individuals in almost all the ages which help them to create and share their personal feelings, experiences, ideas as well as information with others connected to them over a computer mediated technologies Individuals use these social media applications such as Facebook and twitter which are popular most to share their experiences, opinions, day today activities as well as achievements. Due to this nature when there are emergencies and natural disasters these social media applications tend to be flooded with content generated from public who affected, who are looking for their family members and friends, who are looking for information as well as with the people engage in humanitarian activities. Therefore social media has become the first to generate related information when there is a catastrophic event before any of news sites or government bodies engage in disaster management. These social media content is quick accurate and subjective during disaster situations therefore we can use this information as an asset to reduce risk and build awareness among public about the disaster as well as to provide decision making support to relief efforts. This research focuses on building decision making support using social media content generated during disaster situations in Sri Lankan context. Mainly the content will be tweets posted by public during a natural disaster and consisting with text written in English. Therefore situational awareness building will be done using text mining which natural language processing in this study since the content is unstructured. Content will be analyzed using techniques to scrape, clean, classify and generate real information about the disaster and to visualize them to support decision making for authorities engage in disaster management as well as volunteers engage in relief efforts.