Location-Based Sentiment Analysis using Bayesian Networks on COVID-19 Twitter Data
M. Caruso · 2023
The onset of the COVID-19 pandemic and the various variants within it have caused mass public opinion, and many turn to web services and social media platforms to easily and quickly voice these opinions.Of these services and platforms, the microblogging platform Twitter has been widely used by people around the world to voice their opinions on the COVID-19 pandemic.Twitter poses unique challenges which stem from the 280 character limit imposed on a given tweet.Further, advances in technology allow for a precise location to be determined upon the submission of a tweet.This thesis deploys the tree augmented naïve Bayes (TAN) classifier sentiment analysis technique to analyze the sentiment towards and between the B.1.1.529BA.1 Omicron and B.1.617.2Delta variants of the COVID-19 pandemic across provinces and territories within Canada.It was believed that the sentiment would vary across Canadian provinces/territories within a given variant due to provincial/territorial COVID-19 measures and restrictions, and would vary across variants due to the di erence in severity and transmissibility of each variant, however it was observed that sentiment varied across provinces/territories for only some variants and varied across variants for only some provinces/territories. I dedicate this thesis to my dogs, Maya and Pippa. Every day you bring me happiness, love, and comfort.