Sentimen Analisis Berbasis Aspek terhadap Review Aplikasi Digital Wallet Menggunakan Metode Extreme Gradient Boosting

Abstract

The digital wallet application plays a crucial role in the development of modern financial systems. LinkAja is among the digital wallet applications widely used by the community. Public evaluation of this application significantly influences general perceptions and the identification of key aspects affecting user satisfaction. It is imperative for companies to understand user perspectives on the LinkAja application to serve as an evaluation basis for enhancing its performance. Aspect-based sentiment analysis is an effective tool for comprehending user opinions from diverse reviews. This research aims to develop a sentiment analysis model focusing on aspects within user reviews of digital wallet applications. The Extreme Gradient Boosting (XGBoost) method was chosen for its proficiency in addressing classification problems and handling large datasets. The author utilized 2000 user review data extracted from Google Play Store scraping for this study. These 2000 data points will be split into training and testing data in a 70:30 ratio. Each dataset will undergo preprocessing stages to ensure data cleanliness. Subsequently, the author will extract aspects from the training data using the Latent Dirichlet Allocation (LDA) algorithm. Once aspects are identified, words from the training data will be transformed into vectors using Word2Vec features. Following this, aspect-based sentiment analysis classification will be conducted using the Extreme Gradient Boosting method to generate the XGBoost model. The model will then be tested on the testing data, and the evaluation results will be presented in the form of a confusion matrix. The average accuracy results based on four aspects are found to be 90%.