Hybrid Deep Neural Network G-LSTM for Sentiment Analysis on Twitter: A Novel Approach to Disaster Management

Abstract

As it provides real-time insights into public opinion & emotional responses during disasters, social media sentiment analysis has gained in importance in disaster management. Twitter data can be used for sentiment analysis in the field of disaster management by employing G-LSTM, a hybrid deep neural network of Gated Recurrent Units (GRUs) and Long Short-Term Memory (LSTM) networks. Thanks to microblogging networks like Twitter,information pours in quickly after a calamity. Tweets are a good source of data. During acrisis, not all tweets are meant to help. Twitter updates on the death toll, property damage,and actions taken by civilised institutions are all welcome. A great deal of study has beendevoted to the importance of rapidly disseminating disaster-related updates. Studying howpeople use Twitter during calamities like storms and floods could save needless loss of life.data from Twitter's API that has been processed using Natural Language Processing (NLP)in order to anticipate social media disasters Twitter conversations about recent calamities,both real and imagined. Pre-processing includes access to a database of 10,000 manually-classified tweets; N-gram analysis; removal of punctuation; removal of HTML tags;revision of spelling; and vectorization of data. Some of the deep learning algorithmsproposed in this study are the Long Short-Term Memory (LSTM), the Gated Recurrent Unit(GRU), and the Hybrid LSTM-GRU. The measures utilised for assessment are accuracies,precisions, recall rates, and losses. The accuracy, precision, recall, and loss values of thehybrid GRU-LSTM were 0.9978, 0.9948, 0.9931, and 0.0060, respectively, when comparedto those of the LSTM and the GRU Model.