Text Analytics on Twitter Text-based Public Sentiment for Covid-19 Vaccine: A Machine Learning Approach

Abstract

The global pandemic of the novel Coronavirus in 2019, known by the World Health Organisations (WHO) as Covid-19, has put various governments in a vulnerable situation around the world. For virtually every nation in the world, the effects of the Covid-19 pandemic, previously experienced by the people of China alone, has now become a matter of great concern. This research highlights its impact on the global economy, in addition to the immediate health consequences associated with the Covid-19 pandemic. The study further discussed the use of Text Analytics and Sentiment Analysis in Natural Language Processing (NLP) based on Twitter text to analyse public sentiment and derive insights regarding Covid-19 vaccines in the healthcare domain. Two machine learning algorithms were employed: Support Vector Machine (SVM) and K-Nearest Neighbor (KNN) to classify and evaluate the results. Various pre-processing techniques were adopted to help in detecting the public sentiment based on the three sentiment polarity classes: positive, negative, and neutral. The result of the sentiment class distribution reveals that 31% of the public sentiment regarding Covid-19 vaccines is positive, 22% is negative while the remaining 47% were classified as neutral sentiment. The experimented machine learning algorithms reveals that SVM produced 88% accuracy which surpasses KNN with 78% accuracy.