Sentiment Analysis of Deepseek - AI Assistant App

This study is an attempt to analyze user sentiments and provide some insights and perspectives on the DeepSeek app using multiple machine learning models; namely Naïve Bayes (NB), Support Vector Machine (SVM), K-Nearest Neighbors (KNN), Random Forest (RF), and Decision Tree (DT). Our objective was to identify common themes and opinions from Google Play Store reviews about the app's reception. By evaluating these models based on accuracy, precision, recall, and F1-score. The study findings revealed that RF demonstrated the highest performance, achieving an Accuracy of 90.50%, Precision of 90.56%, Recall of 90.50 %, and F1-score of 90.51 %. Moreover, to identify the most mentioned words, word clouds are created and the most frequent words in the positive and negative reviews are determined. The findings indicated that RF effectively captured sentiment patterns while also highlighting significant linguistic themes related to app usage. Such findings provided a clear insight that is contributing to a broader understanding of app reception across different platforms and can inform future improvements to the DeepSeek app.