Hybrid focused crawling on the Surface and the Dark Web

EURASIP Journal on Information Security

Table 1 Summary of the 11 hyperlink selection methods proposed. The table lists the classifiers involved, the local context constraints (where applied), and the network type where each classifier is applied for each method under consideration

Methods	Classifiers					Local context constraints		Classifiers used per Network Type
Methods	L	P	D	SC_PL	DC_PL	t ₁ < score_l < t ₂	\|sw\| ≤ k	Freenet	Tor	I2P	Surf.
1	x							L	L	L	L
2	x		x					D	D	D	L
3	x		x			x		L, D	L, D	L	L
4	x		x			x		L, D	L, D	L, D	L, D
5	x	x				x		L, P	L, P	L	L
6	x			x		x		L, SC_PL	L, SC_PL	L	L
7	x		x				x	L, D	L, D	L, D	L, D
8	x	x					x	L, P	L, P	L, P	L, P
9				x				SC_PL	SC_PL	SC_PL	SC_PL
10					x			DC_PL	DC_PL	DC_PL	DC_PL
11	x				x			DC_PL	DC_PL	DC_PL	L

L stands for the link-based classifier, P for the parent Web page classifier, D for the destination Web page classifier, SC _PL for the static linear combination classifier, DC _PL for the dynamic linear combination classifier (both for the static and the dynamic combination classifiers the link-based and the parent classifier are combined), Surf for the Surface Web. The comma separated entries for the classifiers used per network type, entail a two-step hyperlink selection strategy, where the second classifier is enabled based on the local context constraints