Analisis Sentimen Berbasis Aspek Terhadap Ulasan Pengguna Aplikasi Pegadaian Digital Menggunakan Algoritma Naïve Bayes

Abstract

Pegadaian. PT. Pegadaian's form of transformation is the launch of Pegadaian Digital application. The application aims to facilitate the community and improve the service of products owned by PT. Pegadaian. Based on the monitoring as of 20 October 2022, the Pegadaian Digital application received 3.5 points on a scale of 5. This score is low because it contains many negative reviews. Therefore, it is necessary to analyse the review section of the application to increase the score. The method that can be used to analyse it is aspect-based sentiment analysis. Aspects are those that relate to the experience felt by users, namely aspects of learnability, efficiency, errors, and satisfaction. Sentiment analysis requires an optimal algorithm, one of which is Naïve Bayes. This algorithm was chosen because it is known as a simple but efficient algorithm when processing large amounts of data. This research uses two test scenarios, the first scenario using different ratios and base parameters and the second scenario using the addition of smoothing parameters. The result of this research is that the model with the ratio of 80:20 and the addition of smoothing is the best model for sentiment analysis because it produces the best performance value, with an accuracy value of 92%, precision of 80%, recall of 70% and f1-score of 73%.