Sentiment Analysis of E-Commerc Brand Review Using Multinomial Text Naïve Bayes

Abstract

Technological developments has impact on buying and selling transactions online or e-commerce. Bukalapak is the most popular e-commerce content which used for buying and selling and has a mobile phone and website application, which gives users access to provide sentiment analysis reviews can be used to monitor consumer reviews of their products. Naive Bayes classifier is the one of classification method. Naïve Bayes classifier can be divided into two, namely multivariate Bernouli and Naïve Bayes multinomial text. Multinomial text Naïve Bayes able to reduce errors in document classification. The results obtained by the review given by Bukalapak users on Google Play using the Multinomial Naïve Bayes show that 89% of the reviews rated positively with the prediction accuracy of Naïve Bayes by 91.95%. As much as 80.71% of reviews give a rating of 5. However, there are some negative reviews that are mostly written, namely  “Lambat“, “Ribet“, “Mengecewakan“, “Susah“, and “Perbaiki” it can be used to reference material for Bukalapak in improving the services.

Keywords: E-commerce, Multinomial Naïve Bayes,  Text Mining,  Sentiment Analysis