| Sambara application is an innovation from Bapenda West Java for motor vehicle tax services. The Sambara application expected can be provide efficiency, effectiveness and service improvement. The success of the application can be determined by conducting a sentiment review analysis. Sentiment analysis aims to detect polarity in the text in the form of negative or positive opinions, using text mining. At the text processing stage, the Word N-Gram feature is added as a word identification approach and for classification it uses the Support Vector Machine (SVM) method. This study aims to determine the application of Word N-Gram, the results of the accuracy value using the SVM method, and find out how much influence the application of Word N-Gram on the accuracy value. The highest accuracy value in this research was 89.00% with AUC value of 0.944 (excellent classification) on the amount of data 900, but when uses Bi-gram and Tri-gram results in a decrease in accuracy. The accuracy value with the highest increase is in the application of tri-grams with the amount of 1,200 data. Increase in accuracy value by 0.92% compared to Uni-Gram to 88.59% with AUC value of 0.95. |
