<![CDATA[In the realm of disaster management, the real-time analysis of social media data, particularly from Twitter, has become indispensable. This study investigates the efficacy of various machine learning models in classifying tweets pertaining to disaster scenarios, with the goal of bolstering emergency response systems. A dataset of tweets, categorized as related or unrelated to disasters, underwent a rigorous preprocessing regimen to facilitate the evaluation of five distinct machine learning models: Naïve Bayes, Random Forest, Logistic Regression, Support Vector Machines (SVM), and Long Short-Term Memory (LSTM) networks. The performance of these models was assessed based on accuracy, precision, recall, and F1 score. The results indicated that the SVM model excelled, achieving an accuracy of 89%, precision of 88%, recall of 89%, and an F1 score of 88%, making it the most robust for text classification tasks within the context of disaster-related data. The LSTM model also performed notably well, with an accuracy of 87%, precision of 86%, recall of 87%, and F1 score of 86%, underscoring the potential of deep learning models in processing sequential data. In comparison, Naïve Bayes, Random Forest, and Logistic Regression models demonstrated moderate performance, with accuracy and F1 scores in the range of 76-77% and 72-73%, respectively. These insights are crucial for the development of advanced social media monitoring tools that can significantly enhance the timeliness and precision of crisis response. The research not only highlights the necessity of selecting appropriate machine learning models for specific NLP tasks but also sets the stage for future investigations into the integration of hybrid analytical frameworks. This study establishes a foundation for leveraging machine learning to transform social media data into actionable intelligence, thereby contributing to more effective disaster management and community safety strategies.]]>
Copyrights © 2024
