Detecting Web Spam Based on Novel Features from Web Page Source Code

Table 4

Results of using different 3 feature sets on random forest model.

Feature set	Accuracy	Precision	Recall	F1 score	AUC
Selected existing features	0.911	0.909	0.911	0.909	0.937
Novel features	0.911	0.910	0.911	0.907	0.917
Selected existing + novel features	0.930	0.929	0.930	0.829	0.957