Algorithm Application Support Vector Machine with Genetic Algorithm Optimization Technique for Selection Features for the Analysis of Sentiment on Twitter

Twitter has become one of the most popular micro-blogging platform, recently. Millions of users can share their thoughts and opinions about various aspects and activites. Therefore, twitter considered as a rich source of information for decision-making and sentiment analysis. In this case, the sentiment is aimed to overcome the problem of automatically classifying user tweets into positive opinion and negative opinion. The classifier Support Vector Machine (SVM) used in this study is a machine learning technique that is popular text classifiers, as Support Vector Machine (SVM) algorithm is one that has a linear calcification of the main principles for determining the linear separator in the search space that can best separate the two classes different. But the Support Vector Machine (SVM) has the disadvantage that the appropriate parameter selection problem. The tendency in recent years is to simultaneously optimize the features and parameters for Support Vector Machine (SVM), so as to improve the accuracy of classification on Support Vector Machine (SVM). Genetic Algorithm has the potential to produce better features and becomes optimal parameters at the same time. This research generate text classification in the form of positive and negative tweets on twitter. Measurement accuracy is based on Support Vector Machine (SVM) before and after using a Genetic Algorithm. Evaluation was performed using 10 fold cross validation while accuracy is measured by the confusion matrix and ROC curves. The results of the study showed an increase in accuracy of Support Vector Machine (SVM) from 63.50% to 93.50%.