From: On the development of an information system for monitoring user opinion and its role for the public
Study | Specifications | Results |
---|---|---|
Our research | The first experiment explored the development of ML models. Totally, 132,523 texts on various topics, including Covid-19, were gathered | The best results of accuracy were achieved by DT (0.91–0.95) and RF (0.96–0.99) with the Random oversampling technique |
Akpatsa et al. [16] | This paper analyzed topics, discussions, and concerns about Covid-19 vaccination using Twitter datasets. The final dataset contains 15,239 unique tweets | It achieved the following accuracies with an LR (0.83), an RF (0.83), an SVM (0.84), and an NB (0.77) |
Yeasmin et al. [71] | This research explored Twitter datasets to analyze sentiments on the Covid-19 topic. The dataset included tweets from different states of the USA for 15Â days. A total number of 832,528 tweets were gathered | The following results of classification were achieved with ML algorithms: an LR (0.91), an SVM (0.94), an NB (0.91), k-NN (0.90), a DT (0.96), a RF (0.97), and XGBoost (0.83) |
Daradkeh et al. [72] | This paper describes SA of topics related to Covid-19 vaccine misinformation. A corpus of 40,359 tweets has been collected for the dates between January 2021 and March 2021 | It got the following values of accuracy: a DT (0.81), an SVM (0.78), a k-NN (0.76), and an NB (0.74) |
Mishra et al. [73] | This research paper analyzed the public’s sentiments towards the Covid-19 vaccination in India. The dataset included 5977 tweets before the second wave and 42,936 tweets after the second wave | The following values of accuracy were achieved: an LR (0.61), a DT (0.45), a k-NN (0.58), an RF (0.59), and an XGBoost (0.54) |
Iwendi et al. [74] | This paper focuses on gathering real and fake news data on the topics related to Covid-19. The dataset consisted of 586 true news and 578 fake news and 1100 news articles and social media posts regarding Covid-19 | The ML algorithms achieved the following values of accuracy: a k-NN (0.69), a DT (0.77), and AdaBoost (0.83) |