<![CDATA[Sentiment analysis is a field of natural language processing (NLP) that aims to classify emotions or opinions contained in a text. When training a machine learning model for sentiment analysis, a problem commonly encountered is imbalanced datasets or datasets with uneven class distributions. This study investigates Back Translation’s effect on improving machine learning performance using an imbalanced dataset. The imbalanced dataset to be used is the NusaX Sentiment Analysis dataset. Experiment results show that Support Vector Machine (SVM) models give notable improvement in scores, especially with Back Translation using Javanese as the intermediate language, which provides the best F1 macro score improvement of 1.89% and the best F1 weighted score improvement of 1.52%. On the other hand, Naive Bayes models do not show any notable improvements. The findings indicate Back Translation can adjust class distribution and can boost certain models' sentiment analysis accuracy.]]>
Copyrights © 2025
