Next-Generation Disaster Preparedness: Advanced Text Classification with Crowdsourced Data

Abstract

Effective disaster preparedness is crucial in minimizing the impact of natural and human-made disasters, particularly when early warnings can be generated from crowdsourced data. This paper, titled "Next-Generation Disaster Preparedness: Advanced Text Classification with Crowdsourced Data," leverages advanced text classification techniques to analyze real-time social media streams for disaster prediction. By utilizing GloVe embeddings in combination with the XGBoost algorithm, the system is designed to classify disaster-related messages and provide actionable insights for early disaster detection and response. The model achieved a classification accuracy of 80% and an F1-score of 0.77, reflecting significant advancements in capturing relevant disaster-related messages accurately. Despite initial challenges in handling the noisy nature of social media data, the system demonstrated effective disaster classification after hyperparameter tuning. This study emphasizes the potential of integrating advanced machine learning models with crowdsourced data to enhance disaster preparedness and enable quicker responses, contributing to improved disaster management and mitigation strategies.