AI Summary & Key Takeaways
AI InsightThe classification results for each negative, neutral and positive class were compared separately in our study.
📌 3 Key Takeaways
Abstract
With the emergence of Web 2.0, internet users share their feelings, thoughts and ideas with other people using social networks. Understanding people's thought analysis is important for examining marketing and user feedback in social networks. For this reason, sentiment analysis on social networks with machine learning algorithms is a popular field of study. Our study is based on thesentiment analysis of people against the new coronavirus, which affects the world. People can have different moods due to pandemia. The governance of mental issues must be observed to manage the pandemic time period more successfully. In this article, we retrieved 387,953 tweets due to the ten most frequently used COVID-19 related keywords. The most frequently used keywords about COVID-19 which enable to obtain and assess the reaction of Twitter users are investigated. Even if COVID-19 is a health issue and tweets about COVID-19 is expected to contain negative content, we found positive, negative and neutral tweets to analyze texts using sentiment analysis and machine learning approaches. We applied four classifiers like logistic regression, multinomial naive Bayes, support vector machines and decision tree. These classifiers are well studied and utilized in many studies which we mentioned in our study. The performance of the support vector machine, decision tree and logistic regression classifiers are close to each other. The lowest F-score is obtained from multinominal naive Bayes classifier. The classification results for each negative, neutral and positive class were compared separately in our study.