Vader Meets Multilingual Voices: Klasifikasi Sentimen Ulasan Pada Aplikasi Babble Dengan Bantuan Deep Translator

Abstract

 The rapid advancement of technology today greatly facilitates our access to information—within 
seconds, we can obtain whatever information we need. This also makes it easier to learn various languages 
around the world, as seen in the Babbel application. This study aims to identify sentiment in user reviews of 
the Babbel app by utilizing a combination of Deep Translator, VADER (Valence Aware Dictionary and 
Sentiment Reasoner), and Logistic Regression. User reviews were collected from the Google Play Store, 
resulting in 1,000 multilingual reviews. All reviews in different languages were translated into English using 
Deep Translator. After translation, sentiment labeling was performed using VADER. Then, the text data 
were transformed into numerical form using TF-IDF vectorization. After all these steps, the classification 
process was carried out using a Machine Learning model, namely Logistic Regression. The evaluation phase 
used a Confusion Matrix, and the sentiment classification achieved an accuracy of 89%. This study 
concludes that the combination of lexical-based analysis and machine learning can provide reliable results for multilingual sentiment analysis. In the future, this approach can be further developed by evaluating the 
performance of other classification algorithms.