Analisis Sentimen Ulasan Pengguna Generative AI Menggunakan Naïve Bayes Berbasis Python

Rizky Nurhasanah (STIKOM Tunas Bangsa)
Victor Asido Elyakim (STIKOM Tunas Bangsa)

Jurnal Sains dan Teknologi
Vol. 5 No. 3 (2026): Juni 2026

The advancement of Generative Artificial Intelligence (Generative AI) technology has driven increased use of AI-based applications across various sectors, including education, productivity, digital services, and information retrieval. The high intensity of usage generates massive volumes of user reviews on digital platforms, containing valuable information regarding user satisfaction, experience, and perception of system service quality. This study aims to analyze the sentiment of user reviews toward Generative AI applications using a Python-based Naïve Bayes algorithm. The dataset consists of 50,000 user reviews covering five Generative AI applications: ChatGPT, Microsoft Copilot, Google Gemini, Perplexity, and Claude. The research pipeline includes text preprocessing stages of case folding, cleaning, tokenizing, stopword removal, and feature transformation using TF-IDF with unigram and bigram representations. Data are then classified into three sentiment categories (positive, negative, neutral) using three variants of the Naïve Bayes algorithm: Multinomial, Complement, and Bernoulli. Results show a sentiment distribution of 33,695 positive (67.4%), 13,275 negative (26.6%), and 3,030 neutral (6.1%) reviews. Multinomial Naïve Bayes achieved the best performance with an accuracy of 83.07%, precision 77.79%, recall 83.07%, and F1-score 80.29%. Five-fold cross-validation confirmed stable performance with a mean accuracy of 82.31% and a standard deviation of 0.23%.