Sentiment Analysis to Determine Best Seller for the Same Products on the Tokopedia E-Commerce Portal

Abstract

Tokopedia, as one of the most popular Electronic Commerce portals, has increased people's
interest in shopping online. Tokopedia has made the growth in the number of sellers and buyers on Tokopedia
increase. This number of sellers has an impact on the number of similar products sold by many different
sellers. Tokopedia creates difficulties for prospective buyers because they are confused about determining the
best seller when buying the desired security. Based on these problems, a system is designed to assist decision
making in choosing products from best sellers. The course will generate positive sentiment data on each
product at several different sellers. The method used to create this data is the Text Mining method. This
sentiment data is based on user reviews on the same product from several different sellers. Data is retrieved
using the Web Scraping method on the product detail page on the Tokopedia website. The data taken is
information from a product that has been selected, including user reviews data in it. The data that has been
accepted will go through the Text Preprocessing stages until finally, the information is ready for sentiment
processing. The system will process data by classifying positive and negative sentiments from the user review
data. This classification process uses the Naive Bayes Classifier algorithm. The result of the system is the
sentiment classification and value of positive sentiments on each user review on a product.