Multiclass sentiment analysis on COVID-19-related tweets using deep learning models.

Area of Science:

• Natural Language Processing
• Computational Social Science
• Machine Learning

Background:

• The COVID-19 pandemic, declared in March 2020, spurred a significant increase in social media discourse.
• Understanding public sentiment during global health crises is crucial for effective communication and response.

Purpose of the Study:

• To classify user sentiment expressed in Twitter posts related to COVID-19.
• To evaluate the performance of deep learning models, specifically LSTM networks, in sentiment analysis of pandemic-related tweets.

Main Methods:

• Utilized natural language processing techniques for linguistic analysis of Twitter data.
• Implemented and compared seven deep learning models based on Long Short-Term Memory (LSTM) neural networks.
• Trained models to categorize tweets into negative, neutral, and positive sentiment classes.

Main Results:

• Deep learning models, particularly LSTM-based architectures, demonstrated effectiveness in sentiment classification of COVID-19 tweets.
• Comparative analysis highlighted the strengths of deep learning approaches over traditional machine learning classifiers for this task.
• The study successfully categorized a large volume of tweets reflecting diverse public opinions during the early pandemic.

Conclusions:

• Deep learning models offer a robust method for analyzing public sentiment on social media during health emergencies like COVID-19.
• LSTM networks provide a powerful tool for understanding nuanced public reactions to global events.
• The findings underscore the importance of social media monitoring for public health insights.