Multinomial Nave Bayes Classification Model for Sentiment Analysis of Non-Fungible Token

Abstract

Recently, the development of science and technology has been increasingly advanced. Over time, a technology called Blockchain has become known. NFT (Non-Fungible Token) is a digital asset currently popular in the crypto world. It is a blockchain-based technology. This report presents the results of research conducted on the “Naive Bayes Multinomial Classification Model for Sentiment Analysis related to NFT” using 7060 Twitter data based on the hashtag #NFT. Then preprocessing was carried out, and the labeling process using Vader showed that the sentiment of Twitter users related to NFT tended to be found in 3840 negative sentiments, while 3220 positive sentiments were visualized using a word cloud. After being classified using the MNB method, the data obtained an accuracy of 84%. After evaluating the performance of the model, the results obtained were 84% precision, 81% recall, and 83% f1-score. To analyze the performance of the model, it is measured using a confusion matrix to get TruePositive (830), TrueNegative (661), FalsePositive (122), and FalseNegative (152) results so that the Multinomial Naïve Bayes model can work well in analyzing sentiment