S/n | Author | Title | Methodology | Dataset | Results | Limitation |
---|---|---|---|---|---|---|
1 | Samuel et al. [14] | COVID-19 Public Sentiment Insights and Machine Learning for Tweets Classification | Descriptive textual analytics Positive sentiments were assigned 1, negative sentiment was assigned 0 | Public reliance on social media Tweets | The study explores the viability of machine learning classification methods | There was sufficient directional support for Naïve Bayes and Logistic classification for short to medium length tweets |
2 | Bania [15] | COVID-19 Public Tweets Sentiment Analysis using TF-IDF and Inductive Learning Models | Textual analytic, natural language processing (NLP) and use of artificial intelligence (AI) techniques | 40,000 tweets collected manually from twitter site between 3/07/2020 and 11/07/2020 | Experimental results suggest that Random Forest and Bernoulli’s Naïve Bayes models performed better than the other two classifier models | Twitter data alone may not be sufficient to reflect the general mass sentiments for a nation or for the states |
3 | Shorten et al. [16] | Deep Learning applications for COVID-19 | language modelling: Natural Language Processing, Computer Vision, Life Sciences, and Epidemiology | Nil | Result is based under the true positive, false positive rate curve (AUC), and precision-recall | Interpretability, Generalization Metrics, Learning from Limited Labelled Data, and Data Privacy |
4 | Alanezi and Hewahi [17] | Tweets Sentiment Analysis During COVID-19 Pandemic | k-means clustering and MiniBatch k-means clustering using COVID-19 tweets | WHO data are generated using R Language and have 3500 tweets. Bahrain ministry of health dataset is collected using R Language and has 3500 tweets. The English tweets dataset is having 34 attributes and 23,490 instances. The Arabic tweets dataset is having 34 attributes and 13,088 instances | The highest number of words are classified as neutral with 7184 as 43.2%, then positive words with 4870 as 29.3%. The negative words with 4572 as 27.5% | There are a lot of spam tweets in Arabic with commercial tweets as advertisements that affect the quality of the data and take more time in cleaning |
5 | Ramírez-Sáyago [18] | Sentiment analysis from Twitter data Regarding the COVID-19 Pandemic | NLP tasks and developing classifiers based on algorithms such as Naïve Bayes and Decision Trees | 78,000 tweets that contain selected phrases | Fear is the most common emotion, followed by surprise, sadness, happiness, and anger | Validation of API request by Twitter and the delay in accepting the request by Twitter |
6 | Petersen and Gerken [19] | COVID-19: An exploratory investigation of hashtag usage on Twitter | The data were analysed using data science and natural language processing libraries. Qualitative analysis was performed using thematic analysis | A total of 28.5 M tweets have been retrieved, of which 6.9 M tweets included hashtags | The top three themes regarding the number of hashtags used were related to COVID-19, identifying information, interventions, and geographical tagging | 907 k different hashtags were used. Of these, only 1192 hashtags were used more than 1000 times. The qualitative analysis resulted in 13 themes |
7 | Shi et al. [20] | Social Bots’ Sentiment Engagement in Health Emergencies: A Topic-Based Analysis of the COVID-19 Pandemic Discussions on Twitter | The samples were analysed with Linguistic Inquiry and Word Count (or LIWC) This automated text analysis application extracts structural and psychological observations from text records | This study used a self-designed Python script to collect tweets and authors | A total of 118,720 unique users contributed to these tweets, of which 10,865 (9.15%) were social bots, 107,478 (90.53%) were humans, and 375 (0.32%) were unknown users | It is challenging for researchers to demonstrate the political, social, or economic motivation of social bots. we cannot conclude the intention of social bots based on their sentiment expressions |
8 | Dubey [21] | Twitter Sentiment Analysis during COVID-19 Pandemic | The data are being analysed using “RStudio” and are presented with the help of a cloud, graphs and tables | The data are taken from Twitter (#Covid19#Coronavirus) and total of 10,000 tweets are into consideration to find out the emotion, state and sentiments of people | The top emotion which the people are showing is positive as the score is 62 and are hoping that everything would get well soon and they will return to normalcy which is a good sign for the world | Nil |
9 | Chakrabortyet al. [22] | Sentiment Analysis of COVID-19 tweets by Deep Learning Classifiers- A study to show how popularity is affecting accuracy in social media | The claims have been validated through a proposed model using deep learning classifiers with admissible accuracy up to 81% and fuzzy logic for taming the fuzziness of sentiments | a dataset containing 226,668 tweets collected within a time frame and have been analysed which contrastingly show that there were a maximum number of positive and neutral tweets tweeted by netizens | The validation of this research was proposed using deep learning classifiers with an accuracy of 81% | The research demonstrates that though people have tweeted mostly positive regarding COVID-19, yet citizens were busy engrossed in re-tweeting the negative tweets and that no useful words could be found in WordCloud or computations using word frequency in tweets |
10 | Abdulaziz et al. [7] | Topic-based Sentiment Analysis for COVID-19 tweets | The Latent Dirichlet Allocation (LDA) was adopted for topics extraction, whereas a lexicon-based approach was adopted for sentiment analysis | There were 636,798,623 tweets dataset which was decreased to approximately 600,000 tweets | The findings showed 91% accuracy of tweets with Naïve Bayes while 74% with the logistic regression classification | The dataset contains tweet-ID only; therefore, it got more time to rehydrate it and then extract all information of tweets using the code |