Exploring the Performance of Ensemble Machine Learning Classifiers for Sentiment Analysis of COVID-19 Tweets

Rahman, M.M. Hafizur; Islam, Muhammad Nazrul

doi:10.1007/978-981-16-5157-1_30

Cited by 41 publications

(31 citation statements)

References 30 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Compared to the results in the literature in the accuracy of sentiment classification using social media data via ML techniques, which typically ranges from 70% to 85% [36][37][38][39][40], our classification accuracy rates are among the best, if not the best.…”

Section: Sentiment Classificationmentioning

confidence: 63%

COVID-19 sentiment analysis using college subreddit data

Tian

Liu

2022

PLoS ONE

View full text Add to dashboard Cite

Background The COVID-19 pandemic has affected our society and human well-being in various ways. In this study, we investigate how the pandemic has influenced people’s emotions and psychological states compared to a pre-pandemic period using real-world data from social media. Method We collected Reddit social media data from 2019 (pre-pandemic) and 2020 (pandemic) from the subreddits communities associated with eight universities. We applied the pre-trained Robustly Optimized BERT pre-training approach (RoBERTa) to learn text embedding from the Reddit messages, and leveraged the relational information among posted messages to train a graph attention network (GAT) for sentiment classification. Finally, we applied model stacking to combine the prediction probabilities from RoBERTa and GAT to yield the final classification on sentiment. With the model-predicted sentiment labels on the collected data, we used a generalized linear mixed-effects model to estimate the effects of pandemic and in-person teaching during the pandemic on sentiment. Results The results suggest that the odds of negative sentiments in 2020 (pandemic) were 25.7% higher than the odds in 2019 (pre-pandemic) with a p-value < 0.001; and the odds of negative sentiments associated in-person learning were 48.3% higher than with remote learning in 2020 with a p-value of 0.029. Conclusions Our study results are consistent with the findings in the literature on the negative impacts of the pandemic on people’s emotions and psychological states. Our study contributes to the growing real-world evidence on the various negative impacts of the pandemic on our society; it also provides a good example of using both ML techniques and statistical modeling and inference to make better use of real-world data.

show abstract

Section: Sentiment Classificationmentioning

confidence: 63%

COVID-19 sentiment analysis using college subreddit data

Tian

Liu

2022

PLoS ONE

View full text Add to dashboard Cite

show abstract

“…In this study, three different ensemble Machine Learning models are proposed to classify the data of approximately 12 thousand tweets in the UK into three emotion tags. First, the stacking classifier gave the highest F1 score of 83.5%, while in the second model the voting classifier gave 83.3% and in the last model the bagging classifier gave 83.2% results [11].…”

Section: Related Workmentioning

confidence: 96%

Sentiment analysis with ensemble and machine learning methods in multi-domain datasets

Başarslan

Kayaalp

2023

Turkish Journal of Engineering

View full text Add to dashboard Cite

The first place to get ideas on all the activities considered to occur in everyday life was the comments on the websites. This is an area that deals with these interpretations in the natural language processing, which is a sub-branch of artificial intelligence. Sentiment analysis studies, which is a task of natural language processing are carried out to give people an idea and even guide them with such comments. In this study, sentiment analysis was implemented on public user feedback on websites in two different areas. TripAdvisor dataset includes positive or negative user comments about hotels. And Rotten Tomatoes dataset includes positive (fresh) or negative (rotten) user comments about films. Sentiments analysis on datasets have been carried out by using Word2Vec word embedding model, which learns the vector representations of each word containing the positive or negative meaning of the sentences, and the Term Frequency Inverse Document Frequency text representation model with four machine learning methods (Naïve Bayes-NB, Support Vector Machines-SVM, Logistic Regression-LR, K-Nearest Neighbour-kNN) and two ensemble learning methods (Stacking, Majority Voting-MV). Accuracy and F-measure is used as a performance metric experiments. According to the results, Ensemble learning methods have shown better results than single machine learning algorithms. Among the overall approaches, MV outperformed Stacking.

show abstract

“…Others have investigated the public's opinion and awareness of COVID-19 related events (e.g., protests against lockdown, vaccination, and university reopening) and speeches/comments of political leaders (e.g., Donald Trump) ( Hu et al, 2021 , Jang et al, 2021 ). Current studies have been conducted in several countries such as the U.S. ( Jang et al, 2021 , Lyu et al, 2021 ), the U.K. ( Cheng et al, 2021 , Rahman and Islam, 2022 ), Australia ( Ewing and Vu, 2021 , Wang et al, 2022 ), India ( Barkur and Vibha, 2020 ), China ( Li et al, 2020a , Wang et al, 2020a ), Europe ( Kruspe et al, 2020 ), as well as across multiple countries ( Boon-Itt and Skunkan, 2020 , Matošević and Bevanda, 2020 , Rowe et al, 2021 ). Existing studies focus predominantly on solely English-based content, while a smaller proportion uses either content that is in Chinese and retrieved from Weibo (the largest social media platform in China) ( Li et al, 2020a , Wang et al, 2020a ) or non-verbal content (e.g., emoticons) ( Yamamoto et al, 2014 ); scarce attention has been allotted to sentiment analysis involving multilingual content (discussed in Section 4.1.2 ).…”

Section: Current Progressmentioning

confidence: 99%