Ramadhoni Gibran Maulana, Muhammad Shidiq Budiman, Tegar Hardiansyah Prasetyo, Muhammad Fadhlan Karimuddin, Rizqy Adriansyah, Yayan Hendrian, Shynde Limar Kinanti

Perbandingan Algoritma Klasifikasi Sentimen pada Ulasan Aplikasi Mobile JKN

Introduction

Perbandingan algoritma klasifikasi sentimen pada ulasan aplikasi mobile jkn. Compares SVM, Naive Bayes & KNN for sentiment classification of Mobile JKN app reviews. SVM achieved 90.9% accuracy, proving effective for public service feedback and digital service improvement.

Abstract

Sentiment classification plays an important role in evaluating public response to digital services such as BPJS Kesehatan's Mobile JKN application. This study aims to compare the performance of three machine learning algorithms-Support Vector Machine (SVM), Naive Bayes, and K-Nearest Neighbor (KNN) for classifying user sentiment based on reviews in the Google Play Store. A total of 10,000 user reviews were collected using Python and processed on Google Colab. The research process includes text pre-processing, sentiment labeling based on ratings, data splitting, and model training. Evaluation was conducted using accuracy, precision, recall, F1 score, and confusion matrix metrics. The results show that the SVM algorithm provides the best accuracy of 90.9%, followed by Naive Bayes (90.3%) and KNN (86%). These findings prove that SVM is the most effective model for sentiment classification in the context of public services and provide important insights for government policy evaluation and digital service improvement.

Review

This study presents a timely and relevant comparison of machine learning algorithms for sentiment classification, specifically applied to user reviews of the Mobile JKN application. The research effectively addresses a critical need to evaluate public response to digital government services, a domain with significant societal impact. By systematically comparing Support Vector Machine (SVM), Naive Bayes, and K-Nearest Neighbor (KNN) on a substantial dataset of 10,000 Google Play Store reviews, the authors establish a clear benchmark for sentiment analysis in this context. The finding that SVM achieves the highest accuracy of 90.9% is a significant result, providing valuable insights for improving digital services and informing policy decisions. The methodology employed in this research is robust and well-articulated, demonstrating a clear understanding of the sentiment classification pipeline. The collection of 10,000 real-world user reviews using Python, followed by meticulous pre-processing and sentiment labeling based on ratings, ensures the practical applicability of the findings. The subsequent evaluation using a comprehensive suite of metrics—accuracy, precision, recall, F1 score, and confusion matrix—strengthens the credibility of the comparative analysis. This systematic approach, coupled with the transparent reporting of individual algorithm performance, underscores the study's scientific rigor and its potential to contribute meaningfully to the field of natural language processing applied to public service feedback. While the study provides a solid foundation, future work could further enhance its contributions. A deeper discussion on the specific characteristics of the Mobile JKN review data that might explain SVM's superior performance, perhaps relating to its ability to handle high-dimensional feature spaces, would add valuable theoretical context. Additionally, exploring the application of more advanced deep learning models for sentiment analysis, such as recurrent neural networks (RNNs) or transformer-based models, could offer interesting comparisons and potentially higher performance benchmarks in the future. Nevertheless, this paper makes a significant practical contribution by clearly identifying the most effective traditional machine learning algorithm for sentiment classification in the vital domain of public digital services.

Full Text

Comments

You need to be logged in to post a comment.