Reliable Knowledge Discovery from Social Media Data in Health Crisis in Developing Countries

Abstract

Social media is a barometer to anticipate sentiment of the public about the state of affairs and ongoing pandemic engaged an additional user base who are confined to their stations. COVID-19 startled the world and the crisis exacerbates in the absence of sufficient data for policy making. The data from social media and a timely analysis can provide sufficient statistics for decision-making. This study explores Twitter data to discover knowledgeable statistics on public sentiments about COVID-19 vaccination in developing countries. The study inspects data collected from two extremely populated developing countries: India and Pakistan. Support Vector Machine (SVM) classifier achieves 74.3% accuracy on the manually labeled dataset. Furthermore, the sentiment analysis is correlated with other indigenous factors like regional literacy rate and COVID-19 calamities in the time interval. It is observed that the negative to positive sentiments correlates with a lower to higher regional literacy rate and a higher COVID-19 intensity causes positive sentiments towards vaccination. The correlations of results with indigenous factors may help to advocate the devised strategies to the right audience and social media knowledge discovery with machine learning techniques may help to recover from data scarcity challenges in a medical emergency like COVID-19 in developing countries. Please note: Abbreviations should be introduced at the first mention in the main text – no abbreviations lists. Suggested structure of main text (not enforced) is provided below.

Sentiments Analysis Study for the Adaptation of Online Medical Forums

“…The Tweepy module evaluated two thousand (2000) tweets from Twitter and identified tweets that firmly adopted online medical forums (positive tweets), those that were neutral (neutral tweets) and those that physically opposed the adaptation of online medical systems (negative tweets). The analysed data will be displayed on the bar chart and scatter line graph [6].…”