A Malicious URL Detection Model Based on Convolutional Neural Network

Table 4

Experiment results.

No.	Embedding method	Network structure	Accuracy	F1-score	Recall	Precision
1	Word embedding based on character embedding	Stacked CNN	0.958	0.959	0.976	0.942
2	Word embedding based on character embedding	DCNN	0.961	0.960	0.936	0.984
3	Word embedding based on character embedding	DCNN + extracting fields	0.987	0.987	0.981	0.993