Seminar by Dr Rohitash Chandra
There have been major economic and social consequences due to the closure of businesses and job losses during the COVID-19 pandemic. Psychologists take interest in understanding how people express emotions and sentiments when dealing with catastrophic events. COVID-19 was undoubtedly also a major political issue and coincided with elections in major populated nations such as USA and India. These major events during COVID-19 resulted in active social media usage and harnessing knowledge from the data from platforms such as Twitter can be beneficial for researchers. Advancements in deep learning-based language models have been promising for sentiment analysis and topic modelling. Language models leveraged with data from social networks such as Twitter can provide valuable insights to scientists and policymakers for the management of pandemics. In this seminar, we review some of our recent attempts in sentiment analysis during the rise of novel COVID-19 cases in India. We use LSTM and a pre-trained BERT language model for embedding and sentiment analysis. We first review the sentiments expressed for selective months in 2020 which covers the major peak of novel cases in India. We also implement topic modelling to compare the three major waves in India. Finally, we look at anti-vaxxer sentiments worldwide with the goal of comparing major countries. There are major limitations in accessing data from Twitter and hence we also release a global dataset of major Twitter active countries such as the UK, Brazil, USA, India, Japan, Indonesia, Australia, and Indonesia.
Date: 18th November, 2022
Note: This is a repeat of seminar given at UniMel and UTAS in August 2022
