Analysis of Public Sentiment Toward Crypto Exchange in Indonesia

Analisis Sentimen Publik Terhadap Aplikasi Pertukaran Kripto di Indonesia

Abstract

This research analyzes user sentiment toward cryptocurrency exchange applications in Indonesia using text analysis and Support Vector Machine (SVM). The research stages include review data scraping, preprocessing, sentiment labeling, splitting training and testing data, and model evaluation. The SVM model shows an accuracy of around 80–81%, with stronger performance in recognizing positive sentiment compared to negative sentiment because positive expressions are more explicit. Analysis using word clouds reveals two main narratives: satisfaction with ease of use, security, and application features, as well as complaints about technical issues such as bugs, slow verification, and complicated withdrawal processes. Overall, public perception of both applications tends to be positive. For Indodax, user sentiment is 59.7% positive and 40.3% negative. Meanwhile, for Tokocrypto, 53.8% is positive and 46.2% negative. The researchers also found that positive sentiment is dominated by aspects of ease of use and service reliability. This study emphasizes the importance of combining quantitative and qualitative methods in understanding local user behavior and recommends that developers improve technical performance, strengthen branding, and consider advanced models such as BERT to improve sentiment analysis accuracy.   KEYWORDS: Sentiment Analysis, Support Vector Machine (SVM), Cryptocurrency Applications