“…These features were used to train several machinelearning algorithms for classification, mainly SVM, Multinomial Naïve Bayes (MNB), Conditional Random Fields (CRF), Decision Trees, and k-Nearest Neighbors (k-NN). Overall, in some cases, SVM achieved better results [21,48,57,58,66,90,91,100,101,110,144,145,147,148,175,194,195,205,206,218,237,307,351,363,365], and in other cases, NB performed better [51,115,117,191,257,286], especially in the case of unbalanced datasets such as in References [306,308,309]. Mostafa [304] claimed that the best classifier is dataset dependent.…”