A platform for research: civil engineering, architecture and urbanism
A platform for research: civil engineering, architecture and urbanism
COVID-19 Tweets Analysis Using Hybrid Bi-LSTM and Transformer Models
The COVID-19 pandemic has unleashed a global health crisis, significantly disrupting daily life and affecting individuals physically, mentally, and economically. This unprecedented situation highlights the importance of understanding the emotional impact on mental health, particularly the negative emotions like anger and fear. Analyzing these emotional responses is crucial, as they can lead to significant social repercussions. In this paper, we propose a novel framework to analyze the emotional dynamics in COVID-19 related tweets, aiming to mitigate the adverse effects of the pandemic on mental health by understanding the nuances of emotional responses. Our approach integrates a two-task methodology, leveraging the strengths of Bidirectional Long Short-Term Memory (Bi-LSTM) models and Transformer models, to classify emotions and quantify their intensities, respectively. The first phase of our framework utilizes Bi-LSTMs for a dual-step classification task, enabling precise categorization of tweets into specific emotional and sentiment classes by capturing contextual nuances from both directions of the text sequence. Following classification, we apply Transformer models in a regression analysis phase to accurately quantify the intensities of the identified emotions. Upon evaluating our model, it has outperformed existing models in dual-stage classification, achieving an 84.37% recall, 89.32% precision, 93.12% accuracy, and an 86.18% F1-score. It accurately categorized tweets into emotional categories and sentiments from positive to negative. In regression, it excelled with an R2 of 0.5301, MSE of 0.0037, and RMSE of 0.0611, demonstrating superior prediction of emotion intensities.
COVID-19 Tweets Analysis Using Hybrid Bi-LSTM and Transformer Models
The COVID-19 pandemic has unleashed a global health crisis, significantly disrupting daily life and affecting individuals physically, mentally, and economically. This unprecedented situation highlights the importance of understanding the emotional impact on mental health, particularly the negative emotions like anger and fear. Analyzing these emotional responses is crucial, as they can lead to significant social repercussions. In this paper, we propose a novel framework to analyze the emotional dynamics in COVID-19 related tweets, aiming to mitigate the adverse effects of the pandemic on mental health by understanding the nuances of emotional responses. Our approach integrates a two-task methodology, leveraging the strengths of Bidirectional Long Short-Term Memory (Bi-LSTM) models and Transformer models, to classify emotions and quantify their intensities, respectively. The first phase of our framework utilizes Bi-LSTMs for a dual-step classification task, enabling precise categorization of tweets into specific emotional and sentiment classes by capturing contextual nuances from both directions of the text sequence. Following classification, we apply Transformer models in a regression analysis phase to accurately quantify the intensities of the identified emotions. Upon evaluating our model, it has outperformed existing models in dual-stage classification, achieving an 84.37% recall, 89.32% precision, 93.12% accuracy, and an 86.18% F1-score. It accurately categorized tweets into emotional categories and sentiments from positive to negative. In regression, it excelled with an R2 of 0.5301, MSE of 0.0037, and RMSE of 0.0611, demonstrating superior prediction of emotion intensities.
COVID-19 Tweets Analysis Using Hybrid Bi-LSTM and Transformer Models
Bader, Mohamed (author) / Shahin, Ismail (author) / Ahmed, Abdelfatah (author)
2024-06-03
700697 byte
Conference paper
Electronic Resource
English
UNCOVERING TEMPORAL DIFFERENCES IN COVID-19 TWEETS
| TIBKAT | 2021
Sentiment Analysis of COVID-19 Tweets Using Deep Learning and Lexicon-Based Approaches
| DOAJ | 2023
| DOAJ | 2023
An interpretable hybrid deep learning model for flood forecasting based on Transformer and LSTM
| DOAJ | 2024