<![CDATA[This study focuses on sentiment analysis of user reviews from the top 10 peer-to-peer (P2P) lending applications using Support Vector Machines (SVM) enhanced with linguistic feature augmentation. A total of 73,955 reviews—approximately 60% in Indonesian and 40% in English—were analyzed. The research included descriptive analysis, text preprocessing, word weighting, data labeling, and data visualization. Textual feature augmentation included sentiment polarity scores, part-of-speech (POS) tag frequencies, and domain-specific keywords extracted from a curated corpus to enrich the input space for classification. The SVM model with a Radial Basis Function (RBF) kernel achieved 91.29% accuracy in classifying positive and negative sentiments. Results indicated a predominantly positive perception among users, as reflected in frequently used terms such as “good,” “easy,” and “fast,” suggesting high user satisfaction. However, concerns regarding security and technical issues were also present, captured by terms such as “verification” and “data.” Word cloud visualizations highlighted key sentiment trends. These findings suggest that while public distrust persists in certain areas, positive sentiments significantly outweigh negative concerns. This provides actionable insights for service providers and policymakers to enhance platform reliability, address user pain points, and foster greater trust in P2P lending services.]]>
Copyrights © 2025
