Analisis Sentimen Gemini AI Menggunakan Multinomial Naïve Bayes dengan TF-IDF dan BoW

Abstract

The development of artificial intelligence Large Language Models, such as Gemini AI, has attracted widespread public attention. The advanced development of Gemini AI is inseparable from valuable public reviews for product evaluation, but the massive amount of reviews makes manual analysis inefficient. This study aims to conduct sentiment analysis on Gemini AI application reviews using the Multinomial Naïve Bayes classification algorithm. The primary focus is to compare the performance of two feature extraction methods: Term Frequency-Inverse Document Frequency (TF-IDF) and Bag-of-Words (BoW). A total of 1,000 reviews were collected from the Google Play Store, which, after undergoing preprocessing and data labelling, resulted in 438 data points for analysis. The model evaluation results show that TF-IDF feature extraction provides superior performance with an accuracy of 88% and an F1-Score of 93%, compared to BoW, which produces an accuracy of 84% and an F1-Score of 91%. These results indicate that the TF-IDF feature extraction method is more effective in analysing the sentiment of Gemini AI app reviews using Multinomial Naïve Bayes.