Typhoon-DIG: Distinguishing, Identifying and Geo-Tagging Typhoon-Related Social Media Posts in Taiwan

Typhoons are one of the major hazards in East Asia, including Taiwan, resulting in substantial damage to people and the economy. The development of a real-time hazardous event and damage tracking system is essential for effective typhoon response. While social media offers a valuable platform for sharing disaster information, the challenge of organizing and geotagging posts related to typhoons remains unsolved. In this paper, we introduce Typhoon-DIG, a framework designed for distinguishing, identifying, and geotagging social media posts related to typhoon disasters. Post data is collected from major Taiwanese social media platforms, and it employed the BERT deep learning language model to filter relevant posts and identify geolocation information within the texts for geotagging. Our proposed framework presents practical methods for data collection and organization in disaster tracking system, thereby providing robust support for subsequent disaster assessment and management.