<![CDATA[The development of regulations and trends in electric vehicle adoption in Indonesia has generated diverse public responses and opinions on social media platforms such as YouTube. However, manually evaluating large volumes of unstructured review data tends to be inefficient and less effective. This study applies the Support Vector Machine (SVM) algorithm with a Radial Basis Function (RBF) kernel, utilizing Term Frequency-Inverse Document Frequency (TF-IDF) weighting and optimized preprocessing techniques to automatically classify public opinion sentiment. The dataset consists of 1,517 Indonesian-language YouTube comments that underwent a series of preprocessing stages, including case folding, cleaning, tokenization, stopword removal, normalization, and stemming. The resulting cleaned dataset was then transformed into a numerical matrix containing 1,180 word features and divided into training and testing sets using a 70:30 ratio. The initial labeling results showed that public opinions were dominated by negative sentiment, accounting for 57.3% (869 comments), followed by positive sentiment at 33.2% (504 comments) and neutral sentiment at 9.5% (144 comments). The performance evaluation of the SVM classification model using a confusion matrix achieved an overall accuracy of 78.1%. These findings demonstrate that the application of optimized preprocessing techniques combined with a nonlinear SVM kernel provides reliable performance in accurately identifying and mapping the characteristics of digital public opinion.]]>
Copyrights © 2026
