뉴스 빅데이터 기반 자연재해 전조 감지 분석

Abstract

Identifying early warning signs of natural disasters is critical for mitigating vulnerability and enhancing disaster preparedness. While traditional disaster management often relies on numeric hydrometeorological data, a comprehensive approach should incorporate information on the impacts of disasters on human and environmental systems. This study explores the use of news big data to detect precursors of natural disasters by analyzing disaster-related keywords collected through web crawling from January to December 2022. Focusing on South Korea's 2022 natural disasters—including droughts, heat waves, and severe flooding—the study examines the timing and frequency of disaster-related news mentions across regions. A comparative analysis of news data and actual disaster occurrence timelines highlights the feasibility of using media coverage as a near real-time monitoring tool. Findings suggest that integrating news archives with scientific indices can enhance disaster response systems by improving real-time monitoring and situational awareness. This approach demonstrates the potential for leveraging big data to develop a robust disaster early warning and monitoring system.