Deep learning for COVID-19 topic modelling via Twitter: Alpha, Delta and Omicron

Area of Science:

• Computational linguistics
• Public health informatics
• Social sciences

Background:

• Deep learning methods offer novel approaches to topic modeling.
• Understanding human behavior during extreme events like pandemics is crucial.
• COVID-19 (Coronavirus Disease 2019) has had profound psychological, social, and cultural impacts.

Purpose of the Study:

• To apply advanced deep learning language models for topic modeling of COVID-19 related data in India.
• To analyze the evolution of key themes across different pandemic waves, from the Alpha to the Omicron variant.
• To investigate the correlation between identified topics and prevalent news media during specific periods.

Main Methods:

• Utilized prominent deep learning-based language models for topic modeling.
• Analyzed a dataset encompassing the COVID-19 pandemic's progression in India.
• Examined topic shifts and continuities across distinct variant waves.

Main Results:

• Identified overlapping themes such as governance, vaccination, and pandemic management across COVID-19 waves.
• Discovered emerging novel issues in political, social, and economic spheres during the pandemic.
• Found a strong correlation between major extracted topics and contemporary news media coverage.

Conclusions:

• The deep learning framework effectively captures significant issues during different phases of the COVID-19 pandemic.
• The approach has potential for extension to analyze public discourse in other countries and regions.
• Topic modeling provides valuable insights into societal responses and challenges during global health crises.