Spam Review Classification Using Ensemble of Global and Local Feature Selectors

摘要

In our work, we propose an ensemble of local and global filter-basedfeature selection method to reduce the high dimensionality of feature space andincrease accuracy of spam review classification. These selected features are thenused for training various classifiers for spam detection. Experimental results withfour classifiers on two available datasets of hotel reviews show that the proposedfeature selector improves the performance of spam classification in terms of wellknown performance metrics such as AUC score.