Abstract:
Natural and man-made disasters potentially cause significant damage and disruptionto communities. As population density increases and natural disasters become moreextreme, it becomes increasingly important to both manage communications andextract information from communications to be able to mitigate the negative effectsof such disasters. The emergence of social media platforms has led to new avenuesfor the collection and dissemination of information that is either local or global.Although social studies have revealed that social media feeds can improve effectivedisaster preparedness and recovery, it has been observed that the use of social mediacan have severe negative consequences through the rapid spread of false informationleading to the inappropriate allocation of resources and in extreme cases panic andlawlessness. Rumours and false information are likely to affect appropriate corporateresponses as well as, where appropriate, responses of public organizations taskedwith appropriately responding to natural and man-made disasters. The ability to iden-tify instances of false information through the course of natural and man-made disas-ters is a critical capability for corporate and public bodies in order to improve disastermanagement and response.To reduce the impact of these rumours, a technique is proposed that makes use ofsupervised learning to differentiate between information about an actual event frominformation about a false one and communicating it effectively to appropriate organi-zations. For this purpose, 934 social media feeds were analysed using a Naïve Bayesclassifier. Clearly, applied early on this technique potentially can improve the qualityof disaster response and recovery and mitigate the negative consequences.Broadly speaking, we consider this research to relate to the management of knowl-edge and information flow in disaster situations. Clearly, the techniques introducedin this paper, in providing individuals and organizations with access to knowledgerather than false information and rumours, will help organizations manage resourcesand activities during disasters more efficiently and effectively. The study concludeswith the implications, limitations, and future directions.