Abstract

The importance of sentiment analysis has been widely recognized. Sentiment analysis is effective in evaluating public opinion for decision-making. Sentiment analysis is constructive in extracting opinions and classifying sentiments to determine their polarity. Therefore, sentiment analysis has received considerable research attention from scholars to analyze various news and public information, categorizing it as positive, neutral, or negative. Additionally, research on various sentiment analyses using different technical methods and approaches has become a recent focus of study. That is the reason why this study aims to conduct sentiment analysis and research it using more than one method. This study compares the performance of three sentiment analysis machine learning algorithms: Support Vector Machine (SVM), Naïve Bayes (NB), and Long Short-Term Memory (LSTM). The study results show that SVM achieves the best performance, with an accuracy of 83.60%, an F1-score of 89%, and a recall of 95%. Naïve Bayes follows with an accuracy of 81.20%, precision of 77%, recall of 100%, and F1-score of 87%. LSTM records an accuracy of 81.40%, precision of 87%, recall of 90%, and F1-score of 83%.

Concepts :

Access to Document

SDGs

Citations by Year