<![CDATA[The development of digital technology and the explosion of data on social media have increased the need for accurate sentiment analysis to understand public opinion. This article aims to systematically review the role of data pre-processing techniques and classification algorithms in improving the accuracy of sentiment analysis in social media. Through the Systematic Literature Review (SLR) approach, more than 30 scientific articles from trusted sources were reviewed between 2018 and 2024. The results of the study show that effective pre-processing such as tokenization, stemming, and stop word removal significantly improve the quality of input data, while algorithms such as SVM, Random Forest, and deep learning provide the best performance in sentiment classification. This article is expected to be a conceptual reference for further research and the development of a more precise sentiment analysis system.]]>
Copyrights © 2025
