Identifying Disaster-Related Tweets Using Natural Language Processing and Core Machine Learning Algorithms

Identifying Disaster-Related Tweets Using Natural Language Processing and Core Machine Learning Algorithms

연구 분야: Artificial Intelligence

학회: International Conference on Artificial Intelligence and Speech Technology

In today’s digital era, social media platforms like Twitter are used by people across the globe to share information and updates, such platforms serve as crucial means of communication especially during emergencies. However, the most prominent challenge faced by authorities is the large volume of data and the consequent complexity of extracting crucial information from it. This research aims at finding an optimal solution to tackle the stated problem. The authors used machine learning approaches, such as Naive Bayes and Support Vector Machines (SVM), to identify tweets that needed to be addressed right away. The models used for final evaluation were trained using a dataset of about 8,000 tweets out of which just 3,229 were analyzed to be disaster related. The goal is to accelerate information flow and enable authorities to react to urgent needs more rapidly and efficiently by precisely recognizing these tweets. This method helps with better resource distribution and faster action during emergencies by not just speeding up disaster response but also ensuring that important information isn’t lost in the shuffle. The results of this research indicate that SVM (Support Vector Machine) performs better, which makes it a preferable option for catastrophe prediction based on tweets.