Enchancing K-NN Performance With SMOTE for Sentiment Analysis of Streaming App Reviews

This research aims to analyze the sentiment of user reviews for a popular streaming app on both the Play Store and App Store using the K-Nearest Neighbor (K-NN) method. As the user base expands, reviews increasingly influence app development, guiding improvements and optimizing user experience. However, the large volume of reviews renders manual analysis inefficient and prone to inconsistencies, underscoring the necessity of sentiment analysis to quickly and accurately capture user perceptions. Review data were collected from both platforms, with preprocessing steps such as data cleaning, tokenization, and normalization applied to ensure data consistency. The Synthetic Minority Over-sampling Technique (SMOTE) was used to address class imbalance, enhancing the reliability of classification results. Findings indicate that SMOTE improved model accuracy, raising it from 74% to 82.9% for Play Store data and from 79% to 84.1% for App Store data. Furthermore, a notable difference in sentiment dominance was observed, with positive sentiment prevailing on the Play Store, while negative sentiment was more prevalent on the App Store. These insights reveal that, overall, the app is well received, although certain areas highlighted in negative reviews require further attention to improve user satisfaction.