User-Centric Sentiment Analysis of DeepSeek R1: Leveraging Google Play Store Reviews through Feature-Rich Machine Learning Models

Abstract

In the modern era, the rapid development of AI-based models has made sentiment analysis an essential tool for understanding user perspectives. One such model, Deep Seek (RI version), has gained significant importance in natural language processing tasks. This study conducts a comprehensive sentiment analysis on Deep Seek R1 using four machine learning models—Logistic Regression, Random Forest, Support Vector Machine (SVM), and Naïve Bayes—along with two lexicon-based tools, VADER and Text Blob. Two feature extraction methods, Count Vectorizer and Term Frequency-Inverse Document Frequency (TF-IDF), were employed to evaluate the performance of the machine learning models. The results indicate that Logistic Regression achieves the highest accuracy when using Count Vectorizer, while SVM outperforms other models when using TF-IDF. However, among all approaches, VADER sentiment analysis achieves the highest accuracy at 99%, whereas Text Blob records the lowest accuracy. These findings highlight the effectiveness of lexicon-based approaches in sentiment classification while also demonstrating the impact of feature extraction techniques on machine learning model performance.