<![CDATA[The rapid growth of the online market in Indonesia has changed the business landscape. Tokopedia, one of the leading E-commerce platforms, serves millions of users with a variety of products. In the fierce E-commerce competition, understanding customer reviews is very important. However, performing review analysis manually is a complex and time-consuming task. Sentiment analysis is needed to understand customer preferences, improve service quality, and maintain Tokopedia's competitiveness in the competitive E-commerce market. This study carried out a comparison between three algorithms, Support Vector Machine, Perceptron, and Multinomial Naïve Bayes to evaluate and determine the most effective and accurate algorithm in conducting sentiment analysis of product reviews on Tokopedia. The results of research using 2000 Tokopedia product review data show that Multinomial Naïve Bayes has the highest level of accuracy, reaching 84.00% and precision of 96.00%. Support Vector Machines has an accuracy rate of 80.00% and a precision value of 95.00%. Meanwhile, Perceptron provides 81.00% accuracy and 95.00% precision. Evaluation using the confusion matrix also indicates that Multinomial Naïve Bayes provides superior results with a truth level of 1011 for positive sentiment labels and 860 for negative sentiment labels. This research provides valuable insights regarding sentiment analysis of product reviews on Tokopedia, and the results can be a reference for further research exploring more innovative sentiment analysis methods or the application of technology to increase the efficiency of sentiment analysis in the context of E-commerce.]]>
Copyrights © 2023
