<![CDATA[Threads is a text-based social media application developed by Meta that has gained significant popularity since its launch. However, user reviews on the Google Play Store reveal an imbalanced sentiment distribution, with a dominance of positive sentiment, potentially reducing the accuracy of sentiment classification models. This study aims to evaluate the effectiveness of combining the Extreme Gradient Boosting (XGBoost) algorithm with the Synthetic Minority Over-sampling Technique (SMOTE) to address the data imbalance in user reviews of the Threads application. The dataset consists of 1,000 user reviews, which underwent preprocessing steps including case folding, cleaning, tokenization, stopword removal, and stemming. The data were then represented using the TF-IDF weighting method and analyzed using XGBoost, both before and after applying SMOTE. Results show that without SMOTE, the model achieved an accuracy of 87.60%, with a low recall for the negative class (0.69). After applying SMOTE, accuracy improved to 97.49%, and recall for the negative class reached 0.99, with balanced F1-scores for both positive and negative classes (0.98 and 0.97, respectively). These findings demonstrate that SMOTE is effective in handling class imbalance and enhancing model performance. In conclusion, the integration of XGBoost and SMOTE significantly improves fairness and accuracy in sentiment classification of app reviews, offering valuable insights for the application of machine learning in user opinion analysis. Future research is recommended to use larger datasets and consider deep learning models such as BERT.]]>
