The purpose of this paper is to analyze the influence of social media in disaster management with specific focus on text categorization, by building models to identify victims and supporters and to compare the performance and application of these models. With the role of the internet being omnipotent since its inception in the late 90’s, web 2.0 that emphasizes user-generated content and inter-operability has dominated every aspect of the technology industry. The internet has been the source of news, entertainment and education for more than twenty years now. The most innovative of these technologies is social media, which has become a popular communication medium. Twitter, Facebook, Instagram, YouTube and other similar services have become part of our everyday life. We use these media to share and gather information. In this technological era, information is wealth. Social media is a rich and valuable source of information about customer experiences, opinion, emerging trends, live news feeds etc. Mining these data can give tremendous insight into business requirements. Based on the field of interest, these data have proven to be a goldmine for business analysts to recognize trends and predict future ones, understand online behaviour, the connection between individuals etc. With every individual connected by social media, one can use this vast wealth of information for a better living environment and livelihood. One such application is disaster management. Broadly social media is used in four ways during a disaster:
Creating awareness and providing updates and live information;
For rescue/rehabilitation operations: building communities, volunteering etc;
For collecting funds;
Monitoring and providing insights into the whole situation.
In spite of tremendous advancement in technology, millions of people lose their lives and property every year due to natural and artificial calamities. Social media platform aids crisis managers in propagating information to wider audiences, interacting with the public, monitoring other social media networks to get a better sense of live updates on ground zero during a crisis, in getting better situational awareness, and in improving collaboration for sharing information during an emergency.
With the growing popularity and influence of social media during disaster management, countries like India are yet to utilize them to their full potential. Urban & semi-urban cities have around 75% of the population who are connected with social media. During the Chennai flood in 2016, it was found that one of the most contributing sources towards rescue and rehabilitation process was the social media (Pradnya, 2015). There were lot of talks and hypes in the media about the rising social media influence and how it connects people and resources during times of crisis. Though India’s literacy rate is 75%, not all educated people use social media on a day to day basis to share and receive information. There is also the question of whether this much-talked-about social media influence prevails among all Indian cities or is it more relevant only in metro cities like Chennai, Bangalore, Mumbai, Delhi etc. To satisfy this curiosity, we consider Ockhi cyclone wherein only the semi-urban parts of India is affected. Will the influence of social media be the same in semi-urban parts of India as the study shows in Chennai? Does the government make use of these media to connect and service people? Are the data shared in the Social media useful and relevant during disasters/crisis? And how can we harness the information available in the social media to identify people who need help and those who volunteer help? The focus of this study is to seek answers to the aforementioned queries.
This paper intends to add evidence to the existing academic contributions on disaster management using social media. It applies the existing knowledge of disaster management, social media data collection, text analysis, classification algorithms, validation and visualization techniques to accomplish it. More research in the related field has been focussed on the urban cities but this study aims to understand the effectiveness and purpose of social media in semi-urban areas of India. With the government propagating the idea of digitalization to every corner of India, this study tries to take a peek at the possibility and the reality behind it. The results of this applied research can serve as additional validation for the text analysis techniques to create a platform with automatic classifiers on real time so that the government can facilitate much quicker and efficient disaster management. It can serve as an alternate helpline for people to reach out to the appropriate authorities in times of need. The results can also be utilised as a reference for different disasters to gain insight on data tailoring and training.