Perbandingan Performa Algoritma Support Vector Machine dan K-Nearest Neighbors terhadap Analisis Sentimen Aplikasi Perpustakaan Digital

Abstract

The increasing use of digital library applications in the era of information technology necessitates a systematic evaluation of user experience. Online user reviews contain valuable information that can be analyzed to measure service effectiveness. This study aims to identify user perceptions through sentiment analysis and compare the performance of the Support Vector Machine (SVM) and K-Nearest Neighbors (KNN) classification algorithms. This applied research collects 6,000 reviews via web scraping from the Google Play Store, focusing on two applications: iPusnas and Gramedia Digital. The data processing begins with text preprocessing to clean the text data, followed by feature extraction using Term Frequency-Inverse Document Frequency (TF-IDF) to identify relevant features before classification with both algorithms. The performance of the algorithms is evaluated using a confusion matrix to measure accuracy based on the number of correct and incorrect predictions. The results show that the SVM algorithm achieves higher accuracy, with 77.78% for iPusnas and 79.34% for Gramedia Digital. In contrast, KNN only achieves 40.06% accuracy for iPusnas and 60.54% for Gramedia Digital. From these results, it can be concluded that SVM outperforms KNN in processing digital library application user reviews. This indicates that SVM is more effective in handling large and complex review data. Based on these findings, it is recommended that digital library application developers utilize SVM in sentiment analysis systems to enhance service quality and user experience. Future research may consider aspect-based approaches or a combination of algorithms to achieve more optimal results in sentiment analysis.