COVID-19 Vaccination-Related Sentiments Analysis: A Case Study Using Worldwide Twitter Dataset

Area of Science:

• Public Health
• Computational Social Science
• Artificial Intelligence

Background:

• The COVID-19 pandemic necessitated rapid vaccine development and deployment.
• Public perception and vaccine hesitancy pose significant challenges to vaccination campaigns.
• Social media platforms are crucial for understanding public discourse and sentiment.

Purpose of the Study:

• To analyze global perceptions and perspectives on COVID-19 vaccination using Twitter data.
• To evaluate the effectiveness of various sentiment analysis techniques for public health discourse.
• To develop and validate an advanced machine learning model for sentiment classification.

Main Methods:

• Utilized a worldwide Twitter dataset related to COVID-19 vaccination.
• Employed natural language processing and machine learning for sentiment analysis.
• Developed and tested an ensemble model, Long Short-Term Memory-Gated Recurrent Neural Network (LSTM-GRNN), combining LSTM, GRU, and RNN.

Main Results:

• TextBlob demonstrated better performance than VADER and AFINN in sentiment analysis.
• The proposed LSTM-GRNN model achieved 95% accuracy in sentiment classification.
• The LSTM-GRNN model outperformed existing machine and deep learning models in analyzing COVID-19 vaccination sentiments.

Conclusions:

• The LSTM-GRNN model is highly effective for sentiment analysis in public health contexts.
• Understanding public perception through social media analysis can inform vaccination policies.
• Accurate sentiment analysis is vital for addressing vaccine hesitancy and promoting public health initiatives.