Intelligent lead-based bidirectional long short term memory for COVID-19 sentiment analysis.

Area of Science:

• Natural Language Processing (NLP)
• Deep Learning
• Social Media Analytics

Background:

• Social media platforms are crucial for information dissemination during global events like the COVID-19 pandemic.
• Analyzing public sentiment from vast amounts of social media data presents significant challenges.
• Natural Language Processing (NLP) and deep learning are essential tools for understanding public emotions during health crises.

Purpose of the Study:

• To develop and evaluate a deep learning mechanism for accurately identifying public sentiment from Twitter data related to COVID-19.
• To improve sentiment analysis by minimizing classifier loss during data learning.

Main Methods:

• Utilized a deep learning approach, specifically an intelligent lead-based Bidirectional Long Short-Term Memory (BiLSTM) network.
• Incorporated an intelligent lead optimization technique to reduce classifier loss and enhance learning accuracy.
• Trained and tested the model on online Twitter data concerning the COVID-19 pandemic.

Main Results:

• The intelligent lead-based BiLSTM model achieved high performance metrics: 96.11% accuracy, 99.22% sensitivity, and 95.35% specificity.
• Demonstrated significant performance improvements of 14.24%, 10.45%, and 26.57% in accuracy, sensitivity, and specificity, respectively, compared to the baseline K-Nearest Neighbors (KNN) technique.
• The intelligent lead optimization effectively reduced learning loss, leading to more accurate sentiment analysis.

Conclusions:

• The proposed intelligent lead-based BiLSTM model offers a robust and accurate method for COVID-19 sentiment analysis on social media.
• This approach provides valuable insights into public opinion and emotional responses during health emergencies.
• The study highlights the potential of advanced deep learning techniques in analyzing large-scale social media data for public health surveillance.