Correlation Analysis of Spatio-temporal Arabic COVID-19 Tweets

Abstract

Since the recent COVID-19 outbreak, several researchers have begun to focus on various difficulties to data mining of social data to study people's reactions to the outbreak. Recent approaches have mostly concentrated on the analysis of social data in the English language. In this study, we present an in-depth social data mining approach to extract Spatio-temporal and semantic insights about the COVID-19 pandemic from the Arabic social data. We developed sentiment-based categorization methods to extract major topics at various location granularities (regions/cities). Besides, we used topic abstraction levels (subtopics and main topics). A correlation-based analysis of Arabic tweets and official health provider data will also be presented. Furthermore, we used occurrence-based and statistical correlation methodologies to create many topic-based analysis mechanisms. Our findings demonstrate a positive association between top subjects (for example, lockdown and vaccine) and the increasing number of COVID-19 new cases, but unfavorable attitudes among Arab Twitter users were generally heightened during this pandemic, on issues such as lockdown, closure, and law enforcement.