Table 3.

Classification results under different feature selection settings. `NER’ denotes the NER step and `BIN’ represents the feature binning step. A ‘+’ sign represents employing the respective selection step, while a ‘–’ denotes its exclusion. Standard deviation is shown in parentheses. The highest performance level along each metric is shown in boldface

Feature selection methods	Precision	Recall	F-measure	MCC
NER-, BIN-	0.721 (0.013)	0.678 (0.006)	0.699 (0.007)	0.652 (0.009)
NER-, BIN+	0.752 (0.004)	0.736 (0.010)	0.744 (0.006)	0.702 (0.007)
Our final classifier (NER+, BIN+)	0.719 (0.008)	0.791 (0.012)	0.753 (0.004)	0.711 (0.004)