Enhanced sentiment analysis regarding COVID-19 news from global channels

Area of Science:

• Computational linguistics
• Public health informatics
• Artificial intelligence in healthcare

Background:

• Effective media coverage of disease-related issues is vital for public health awareness and risk reduction.
• Deep neural networks offer powerful tools for textual sentiment analysis, enabling real-time monitoring of health topics.
• Analyzing public sentiment towards health crises like COVID-19 is crucial for understanding societal response and informing public health strategies.

Discussion:

• The proposed Cov-Att-BiLSTM model integrates attention mechanisms, embedding techniques, and semantic labeling for enhanced sentiment analysis accuracy.
• This novel approach was rigorously evaluated against existing deep and machine learning classifiers, demonstrating superior performance.
• The model's effectiveness was validated through the analysis of a large dataset of COVID-19 related tweets, reflecting global sentiment and news trends.

Key Insights:

• The Cov-Att-BiLSTM model achieved a testing accuracy of 0.931 in sentiment analysis of COVID-19 news headlines.
• The approach accurately captured global sentiment and coverage related to the COVID-19 pandemic and vaccination efforts.
• This demonstrates the potential of advanced deep learning models for real-time public health surveillance and analysis.

Outlook:

• Future research can explore adapting the Cov-Att-BiLSTM model for sentiment analysis of other critical public health issues.
• Further development could involve real-time integration of this model into public health monitoring systems.
• Expanding the model's capabilities to include multilingual sentiment analysis could provide broader global health insights.