Understanding User-Level IP Blocks on the Internet

Table 3

Performance of ablation.

Model	Parameter	Dataset	Precision (%)	Recall (%)	F1 (%)
Node Similarity Model	α = 1, β = 0, γ = 0	Train	69.78	54.62	61.28
		Test	66.42	52.32	58.53
Neighbours Similarity Model	α = 0, β = 1, γ = 0	Train	73.45	50.24	59.67
		Test	70.89	49.80	58.50
Modularity Model	α = 0, β = 0, γ = 1	Train	69.57	59.78	64.30
		Test	65.59	56.54	60.73
Node + Neighbours Similarity Model	α = 0.5, β = 0.5, γ = 0	Train	74.89	59.87	66.54
		Test	70.76	55.42	62.16
Node Similarity + Modularity Model	α = 0.5, β = 0, γ = 0.5	Train	73.98	59.45	65.92
		Test	70.82	55.67	62.34
Neighbours Similarity + Modularity Model	α = 0, β = 0.5, γ = 0.5	Train	67.67	53.45	59.73
		Test	67.19	52.44	58.91
ULIB	α = 0.6, β = 0.2, γ = 0.2	Train	74.22	59.67	66.15
		Test	71.23	62.23	66.43