Enhancing Trustworthiness and Minimising Bias Issues in Leveraging Social Media Data for Disaster Management Response (2409.00004v1)

Abstract: Disaster events often unfold rapidly, necessitating a swift and effective response. Developing action plans, resource allocation, and resolution of help requests in disaster scenarios is time-consuming and complex since disaster-relevant information is often uncertain. Leveraging real-time data can significantly deal with data uncertainty and enhance disaster response efforts. To deal with real-time data uncertainty, social media appeared as an alternative effective source of real-time data as there has been extensive use of social media during and after the disasters. However, it also brings forth challenges regarding trustworthiness and bias in these data. To fully leverage social media data for disaster management, it becomes crucial to mitigate biases that may arise due to specific disaster types or regional contexts. Additionally, the presence of misinformation within social media data raises concerns about the reliability of data sources, potentially impeding actionable insights and leading to improper resource utilization. To overcome these challenges, our research aimed to investigate how to ensure trustworthiness and address biases in social media data. We aim to investigate and identify the factors that can be used to enhance trustworthiness and minimize bias to make an efficient and scalable disaster management system utilizing real-time social media posts, identify disaster-related keywords, and assess the severity of the disaster. By doing so, the integration of real-time social data can improve the speed and accuracy of disaster management systems