A comparison of word embedding-based extraction feature techniques and deep learning models of natural disaster messages classification

Abstract

The research aims to compare the classification performance of natural disaster messages classification from Twitter. The research experiment covers the analysis of three-word embedding-based extraction feature techniques and five different models of deep learning. The word embedding techniques that are used in this experiment are Word2Vec, fastText, and Glove. The experiment uses five deep learning models, namely three models of different dimensions of Convolutional Neural Network (1D CNN, 2D CNN, 3D CNN), Long Short-Term Memory Network (LSTM), and Bidirectional Encoder Representations for Transformer (BERT). The models are tested on four natural disaster messages datasets: earthquakes, floods, forest fires, and hurricanes. Those models are tested for classification performance