Analisis Sentimen pada Ulasan Aplikasi Notion AI dengan Metode Support Vector Machine dan Random Forest

Abstract

In the digital era, the utilization of Artificial Intelligence (AI) has been rapidly expanding across various fields, including information management through applications such as Notion AI. This study aims to analyze user sentiment toward the Notion AI application based on review comments on the Google Play Store using two machine learning algorithms, namely Support Vector Machine (SVM) and Random Forest. The data were obtained via web scraping, comprising 300 review comments, 150 positive and 150 negative. The dataset was then divided into 80% training data and 20% testing data to ensure that the model evaluation was conducted objectively using data that were not involved in the training process. The research process included stages of data collection, preprocessing, classification modeling, model evaluation, data presentation, and analysis using the RapidMiner tool. The results showed that the Random Forest algorithm outperformed SVM, achieving an accuracy of 95.97%, a precision of 98.27%, a recall of 94.34%, and an AUC value of 1.000. Meanwhile, the SVM model produced an accuracy of 85.97% and an AUC of 0.954. This study indicates that Random Forest is more effective in handling variations in text data and provides more accurate classification results. Overall, the majority of user reviews of Notion AI are positive, particularly regarding the ease of AI writing features and productivity enhancement, while negative reviews generally relate to language limitations and paid features.