<![CDATA[The evolution from conventional services to mobile banking has led to increased usage of digital banking applications, including BCA Mobile and myBCA. The large volume of user reviews on the Google Play Store serves as a valuable source of information regarding user experiences and perceptions of application service quality. This study aims to analyze the sentiment distribution of BCA Mobile and myBCA user reviews, compare the performance of the Bidirectional Long Short-Term Memory (BiLSTM) and Support Vector Machine (SVM) methods, and determine the most optimal method for sentiment analysis. The dataset consists of 5,000 user reviews for BCA Mobile and 5,000 for myBCA, collected from the Google Play Store. The research process involved data collection, sentiment labeling, and data preprocessing comprising cleaning, case folding, normalization, tokenization, stopword removal, and stemming followed by modeling using BiLSTM and SVM with a linear kernel. Model evaluation was conducted using a confusion matrix based on four metrics accuracy, precision, recall, and F1-score along with McNemar's test to determine the significance of performance differences between the two methods. The results indicate that negative sentiment dominates both applications: 56.5% (2,741 reviews) for BCA Mobile and 57.6% (2,825 reviews) for myBCA. Model comparisons show that SVM achieved an accuracy of 86% for BCA Mobile and 87% for myBCA, whereas BiLSTM achieved 84.5% and 85.6%, respectively. McNemar's test yielded p-values of 2.60 × 10⁻⁵⁴ for BCA Mobile and 5.84 × 10⁻⁶³ for myBCA; since both p-values were below 0.05, it is concluded that there is a significant difference in performance between the two methods. Based on these findings, SVM is the most optimal method for analyzing user reviews of BCA Mobile and myBCA in this study, and the results can serve as a basis for evaluating and improving application service quality and features.]]>
