Covid-19 fake news sentiment analysis.

Area of Science:

• Artificial Intelligence
• Data Science
• Computational Linguistics

Background:

• COVID-19 misinformation proliferates on social media, posing public health risks.
• Social media platforms have struggled to effectively combat the spread of fake news.
• This research addresses the challenge of identifying and mitigating health misinformation.

Discussion:

• Information Fusion techniques were employed to gather diverse datasets, integrating real news from reputable sources with fake news from social media.
• A comprehensive set of 39 features was extracted from multimedia texts to train deep learning models.
• The study evaluated state-of-the-art deep learning models, including Recurrent Neural Network (RNN), Gated Recurrent Units (GRU), and Long Short-Term Memory (LSTM).

Key Insights:

• The developed deep learning model significantly enhanced fake news detection accuracy from 59.20% to 86.12%.
• High precision of 85% was achieved for fake news detection using the RNN model.
• The GRU model demonstrated the best recall (83%) and F1-Measure (83%) for fake news detection.
• GRU, RNN, and LSTM models achieved excellent performance for real news detection, with precision, recall, and F1-Measure ranging from 88% to 90%.

Outlook:

• The proposed model demonstrates superior performance compared to traditional machine learning algorithms for fake news detection.
• Further research could explore real-time detection and cross-platform analysis of health misinformation.
• This approach holds potential for broader applications in combating misinformation across various domains.