Political Post Classification based on Firefly and XG Boost

Abstract

Opinion mining is used in practically every aspect of human life and has a substantial effect in our behaviors. There is a lot of data that displays people' opinions in many domains, like business and politics, due to the expansion and use of online technology. To generate our vector, we used the firefly method to select the finest words from political Arabic posts and looked into two feature extractions: term frequency and term frequency inverse document frequency. These features utilized with XGBoost algorithm to classified the right class into (Revolutionary, Conservative, andReform). Accuracy, F1-score, recall,precision,and number of correct predict were calculated to measure the applied classifiers' performance.The results expose that the TF conclude the best results in accuracy of 98.052% with length of features 210.