KOMPARASI FITUR SELEKSI PADA ALGORITMA SUPPORT VECTOR MACHINE UNTUK ANALISIS SENTIMEN REVIEW

Abstract

ABSTRAK The main problem in the process sentiment analysis review is how to choose and use the best feature selection to get the maximal result. The accuracy of the use of algorithm in analysis sentiment review also have been an important role in the determination results of the analysis. Analysis of the sentiment is a study computing on an opinion, behavior and emotion of a person to an entity. This research also discussed comparative studies, technique classification and combining method of the feature selection to comparsion result of the people opinion about tourist destination. The classifications technique to analyze sentiment review of the tourist destinations, using support vector machine algorithm (svm) and a model of the features selection will be compared between a particle swarm optimization and genetic algorithm to increase the accuracy classifications of support vector machines algorithm. The measurement of were based on accuracy support vector machines before and after the addition of features. The evaluation uses 10 cross fold validation. While the measurement of accuracy measured by confusion the matrix and a curve roc. The result showed an increase in accuracy support vector machines of 75.33 % to 88.67 %. Keywords: Sentimen Review, Support Vector Machine, Analysis Review, Feature Selection.