Sentiment Analysis of M-Paspor App Reviews Using Multinomial Naive Bayes
Journal of Logistics Informatics and Service Science · 2024
This study conducts sentiment analysis on user reviews of the M-Paspor application, an Indonesian mobile application for passport services, available on the Google Play Store in 2023.By leveraging the Multinomial Naive Bayes algorithm, a machine learning technique suitable for text classification tasks, the research aims to evaluate the sentiment polarity (positive, negative, or neutral) of user reviews.The analysis is based on a dataset of 6,443 reviews scraped from the Google Play Store, spanning the period from January 1, 2023, to October 18, 2023.The study employs a comprehensive methodology, including data preprocessing techniques such as case folding, tokenization, and normalization, as well as feature extraction using TF-IDF (Term Frequency-Inverse Document Frequency).The performance of the Multinomial Naive Bayes model is rigorously evaluated using various metrics, including accuracy, F1-score, precision, and recall, calculated through a confusion matrix.The impact of different train-test data split ratios (60:40, 70:30, and 80:20) on model performance is also investigated.The results indicate that the 80:20 data split yields the highest performance, with an accuracy of 66.45%, an F1-score of 64.79%, a precision of 72.24%, and a recall of 64.12%.This study contributes to the understanding of user sentiments and experiences with the M-Paspor application, providing valuable insights for enhancing user satisfaction and improving the quality of digital government services.