ANALISIS SENTIMEN MENGGUNAKAN ALGORITMA NAÏVE BAYES TERHADAP KOMENTAR APLIKASI TOKOPEDIA

Tokopedia is the online shop most visited by Indonesians, with a total visitor number of 1.2 billion, divided from 863.1 million visitors from the mobile web and 329.8 million visitors from the desktop. Even though it is in the top rank and
rating 5 gets the most value, of course not all of them give positive comments, some of them give negative comments, the element of user trust plays an important role in the continuity of an online store, sometimes products are sold or
purchased through one of the goods e-commerce sites (the product is ) does not match the photo in the advertisement and also sometimes the goods are not what you want. To research and analyze this, we need a method and analysis to classify user comments into several categories, which in this study are positive and negative categories. This study uses the Naive Bayes method to generate positive and negative sentiments towards comments from users of the Tokopedia application on Playstore. From the results of tests carried out on 500 user comment data using the TFIDF process and Term Frequency produces the same value, Testing is based on the accuracy value, Precision and Recall, with an accuracy performance value of 91.20%, with a precision value of 1 While on Recall resulting in a value of 87.68%. AUC value 0.944