<![CDATA[The development of information and communication technology has significantly influenced public services, particularly through the adoption of mobile applications like DUKCAPIL, which simplifies access to population administration services. This study aims to analyze sentiment regarding the application by employing ensemble learning techniques and the SMOTE method to address data imbalance. The Extra Trees algorithm is compared against nine other algorithms, including Random Forest, Gradient Boosting, and LSTM. Extra Trees achieves the highest accuracy of 95.29% and outperforms in precision, recall, and F1-score. Deep learning models showed improved accuracy from 76.34% in the initial epoch to 91.56% in the final epoch. XGBoost and Random Forest also demonstrated strong performances, with accuracies of 90.55% and 92.66%, respectively. The results underline the superiority of Extra Trees in terms of stability and accuracy while highlighting the potential of deep learning for model enhancement. These findings provide valuable insights for the development of mobile application-based public services.]]>
Copyrights © 2025
