Analisis Sentimen Ulasan Aplikasi Indrive Menggunakan Algoritma Support Vector Machine (SVM)

Abstract

Google Play Store is a platform that provides a wide variety of applications and games. The vast selection of available applications requires users to be more selective in choosing the right app. One of the applications available on Google Play Store is InDrive. Users often express their opinions about InDrive through reviews on the platform. These reviews serve as a valuable resource for evaluating application quality and frequently influence users' decisions when selecting an app. However, the information contained in user reviews is highly diverse, making it difficult to determine whether a review expresses a positive or negative sentiment. This issue can be addressed using sentiment analysis. This study employs the Support Vector Machine (SVM) algorithm to perform sentiment analysis on user reviews of the InDrive application. The primary objectives are to evaluate the performance of the SVM algorithm and to analyze user sentiment toward the application. The sentiment analysis process involves preprocessing user reviews, converting them into numerical representations, and classifying them using SVM. The classification process includes selecting the best SVM kernel among four options: linear, radial basis function (RBF), polynomial, and sigmoid. The results indicate that the linear kernel performs best, achieving 96% accuracy in training and 91% accuracy in testing. Additionally, sentiment analysis utilizes opinion mining techniques to identify key features influencing positive and negative sentiments. The findings reveal that high-weight positive words generally reflect aspects appreciated by users, such as excellence in service, ease of use, and sustainability of the application. Conversely, high-weight negative words indicate several issues that need improvement, such as concerns about drivers' trust in the system, particularly regarding fairness, payments, and ratings.