Improved Sentimental Response for Classifying Emergency Incidence through Hybridized Mining Technique

Abstract

This research addresses the classification of emergency incidents arising from both natural and human-induced events, emphasizing the necessity for timely intervention and strategic mitigation. It introduces a hybrid data mining approach that integrates Natural Language Processing (NLP) with Bayesian Belief Learning (BBL) to enhance sentiment analysis during crisis scenarios. Real-time data is extracted from Facebook through the Graph API using Python’s requests library. The collected data undergoes preprocessing and is stored in a MySQL database, while the system interface utilizes XML and PHP to display sentiment outcomes. The integration of supervised learning into the NLP process resulted in a signal precision exceeding 92.8%, surpassing the accuracy of existing approaches. A confusion matrix is employed to assess the model’s performance, confirming its high level of predictive precision. The system demonstrates strong capabilities for improving proactive emergency detection and management.