A Post COVID-19 Analytics of African Users Perception of Online Learning


Zainab Olorunbukademi Abdulkareem
University of Ilorin
Yetunde Folajimi
Wentworth Institute of Technology
Sulyman Age Abdulkareem
University of Ilorin


This study explores the potentials of user-generated text on twitter to offer insights into the pre- and post-COVID'19 attitude on e-learning users in Africa. First, we manually assigned positive, negative or neutral sentiment to each of the 1193 tweets collected based on their content. From the sentiment tagging, we found that half of the tweets posted about e-learning are neutral, 27% are negative and 23% are positive sentiments. Furthermore, we evaluated the tested the predictive accuracy of VADER and TextBlob automatic sentiment assigning Phython Libraries. By comparing the VANDER and TextBlob results to the manually assigned sentiments, we found the accuracy to be low at 51% and 45% respectively. This calls for more research to improve sentiments prediction of African tweets. Thus, we report our findings, including data analytics of the extracted tweets, and our future plans to create a model that would take African slangs and expressions into consideration for better sentiments prediction of African tweets.

February 17, 2024
Online ISSN