EXPLORING FACEBOOK USER RESPONSE TO COVID-19 VACCINATION PROGRAMME WITH SENTIMENT ANALYSIS AND TOPIC MODELLING

Abstract

In response to the COVID-19 pandemic, the Malaysian government has implemented the National COVID-19 Immunisation Programme. However, little research has been published on the perceptions of Malaysians regarding the vaccination programme. This study aims to (1) analyse the sentiment analysis of Facebook user perceptions using a Support Vector Machine and (2) identify the topics associated with the introduction of a COVID-19 vaccination programme in Malaysia for each sentiment. Facepager software was used to retrieve all vaccine-related comments. The R software was subsequently utilised for data cleaning and analysis. A linear Support Vector Machine regression model was used to classify the data into negative, neutral, and positive sentiments through sentiment analysis. To accomplish community detection in semantic network analysis and identify the topics for each sentiment, a clustering technique based on the Louvain method was employed. From a total of 5055 comments, 3269 (64.62%) are categorised as negative, followed by 1068 (21.11%) as positive, and 722 (14.27%) as neutral sentiment. The most discussed topics for negative-, positive- and neutral-sentiment were the negative effects of vaccines (74.9%), concern on vaccine adverse effects (69.0%), and lack of confidence (53.1%), respectively. The study findings can aid the local government and agencies to timely address public concern on COVID-19 vaccines.