Data Mining Technology Application in False Text Information Recognition

Table 4

Classification results for J48, SVM, NN, and Naïve Bayes.

Classifier	Feature set	Accuracy (%)	Recall (%)	F-measure	ROC area	Growth rate (%)
J48	FS1	85.74	85.70	0.857	0.904	—
	FS2	92.15	92.10	0.921	0.935	7.47
	FS3	91.94	91.90	0.919	0.926	−0.22
	FS4	91.74	91.70	0.917	0.926	−0.22
SVM (SMO-RBF)	FS1	85.95	86.00	0.859	0.86	—
	FS2	93.60	93.60	0.936	0.936	8.96
	FS3	94.63	94.60	0.946	0.946	1.07
	FS4	95.04	95.00	0.95	0.95	0.42
NN	FS1	87.60	87.60	0.876	0.952	—
	FS2	92.36	92.40	0.924	0.957	5.48
	FS3	94.21	94.20	0.942	0.979	1.95
	FS4	94.21	94.20	0.942	0.969	0
Bayes	FS1	84.30	84.30	0.842	0.935	—
	FS2	86.78	86.80	0.867	0.953	2.97
	FS3	88.84	88.80	0.888	0.957	2.42
	FS4	88.84	88.80	0.888	0.941	0