<![CDATA[Pertamina has issued a cashless application for fuel purchases since July 2019, named as MyPertamina. The application aims to make it easier for customers to make payments in transactions at fuel stations. MyPertamina application can currently be downloaded on Google Playstore. Since its release until now, MyPertamina has been downloaded as many as 10 million with a rating of 3.1 and 339 thousand reviews. Unfortunately the low rating and user reviews dominated by negative comments show that the app's performance is still not satisfactory. The reviews data can be converted into valuable information by using entiment analysis. Many researchers have applied sentiment analysis to MyPertamina user comment data. However, there have been no studies that apply the handling of negation in MyPertamina reviews, even though negative comments are very often found the word negation, i.e ‘tidak’,’jangan’,’belum’ and ‘bukan’ that will change the sentiment of next sentiment word. Untreated negation words will lead to errors in classification which in turn will decrease accuracy. This study applies the handling of negation words using First Sentiment Word (FSW) and Fixed Window Length (FWL) methods. The classification algorithms used are Naïve Bayes Classifier (NBC) and Support Vector Machine (SVM). In this work, we analized 1000 comments consisting of 390 positive comments and 610 negative comments. The results showed that the best performance of negation handling is FSW. This method has improved accuracy by 2.5% and F1 by 1.5% using NBC algorithm and has improved accuracy by 2.9% nad F1 by 3.4% using SVM algorithm.]]>
