The Comparison of Sentiment Analysis Algorithm for Fake Review Detection of The Leading Online Stores in Indonesia

Online store users continue to grow and develop throughout the year so that the number of reviews from customers is also increasing. Customers can write reviews about the products and services and then post them on social media, whether it's a good or bad review. The reviews written are also very diverse such as product specification review, product advantages, product disadvantages and so on. Sometimes it is difficult for readers to determine which reviews are real and correct because there are so many written reviews. Reviews really play a very important role for potential customers to decide whether to buy the product or service or not. This study used several algorithms of sentiment analysis to detect fake reviews of online stores on Twitter. The algorithms were used to calculate the accuracy value of the review. The algorithms used were support vector machine, Naive Bayes, and logistic regression. The accuracy values of the three algorithms were compared and the highest is selected. The review text that was processed to produce fake review detection comes from the 3 leading online stores in Indonesia namely Tokopedia, Shopee, and Bukalapak. The results of calculating the accuracy value for real data using support vector machine is 70.58%, using naïve bayes is 69.89%, and using logistic regression is 70.49%. Comparison of these three results determines that the most appropriate algorithm for detecting fake reviews of online stores is the support vector machine algorithm.