Sustainable Vocational Innovation in E-Government: Development of a Sentiment Analysis For Evaluating SIREKAP Mobile Performance Using Naive Bayes

Abstract

The Regional Head Elections (Pilkada) represent a crucial moment in the democratic process that demands transparency and accuracy in vote counting. The SIREKAP Mobile application was designed to support efficient and transparent vote recapitulation in Indonesia. However, during the 2024 elections, the application attracted public criticism due to technical issues and its perceived impact on public trust. This study aims to analyze user sentiment toward the SIREKAP Mobile application during the 2024 Tulungagung Regency elections using the Naïve Bayes algorithm. User reviews were collected from the Google Play Store, TikTok comments, and online surveys, and classified into three sentiment categories: positive, neutral, and negative. Preprocessing steps included tokenization, stopword removal, and stemming, followed by sentiment classification using a semi-supervised approach combining Naïve Bayes with Expectation Maximization (EM). Evaluation results showed strong performance with 83.5% accuracy, 83.6% precision, 83.5% recall, and an F1-score of 83.5%. Sentiment classification revealed that 69.8% of feedback was negative, 15.5% positive, and 14.7% neutral. This study provides insights into public perception of electoral technology and contributes to improving the transparency and accountability of future elections through enhanced application design and implementation.