Sentiment Analysis On Covid-19 Outbreak Awareness Using Naïve Bayes Algorithm

Abstract

Sentiment analysis has gained much attention nowadays among the researchers especially during the Covid-19 pandemic. Due to the increasing volume of data coming from the social media platforms, researchers have been using sentiment analysis to analyse topics regarding commercial products, daily issues among the society and also to detect important events from the community. Since the social media users are consisting of the community, content that are shared could also be used to detect possible situational hazard such as the outbreak of Covid-19 in advanced. The result from the sentiment analysis could be beneficial to government organizations in order to contain the outbreaks and public health crisis related to Covid-19. The objective of this research is to explore Naive Bayes algorithm for the sentiment analysis on the Covid-19 outbreak awareness based on Twitter data. In this research, the data were collected during the Malaysia's second lock down, which was between the months of April to June 2021 using the Twitter API Tweepy. After the pre-processing and feature extraction stages, the data have been divided into the training and testing dataset for the Naive Bayes sentiment classification. The result has shown that Naive Bayes has been able to generate high performance with more than 90% accuracy for this classification problem. Future work would include the improvement of data preprocessing, more balance of dataset, enhancement of the algorithm and also comparing the performance with other well-known classification algorithms.