Network-theoretic information extraction quality assessment in the human trafficking domain

Applied Network Science

Table 6 Pearson correlation coefficients between precision and network metrics

Network metric	City	Name	Phone
Order	-0.7292 ^∗	-0.8219 ^∗	-0.4537
Power-law exponent	-0.6995	0.8628 ^∗∗	-0.6169
Num. connected components	0.9741 ^∗∗	0.9624 ^∗∗	0.9886 ^∗∗
Clustering coefficient	0.7010	0.9017 ^∗∗	0.8362 ^∗∗
Degree correlation	0.3245	0.2297	0.157
Order (of largest connected component)	-0.9839 ^∗∗	-0.7776 ^∗	-0.9958 ^∗∗
Algebraic connectivity	0.9837 ^∗∗	-0.5925	0.7055
Vertex connectivity	0.9838 ^∗∗	0.5204	0.6025
Edge connectivity	0.9834 ^∗∗	0.5204	0.7229 ^∗
Diameter	-0.5471	0.8368 ^∗∗	-0.9950 ^∗∗
Avg. shortest path length	-0.1373	0.4637	-0.9964 ^∗∗

Since the network is disconnected, all metrics listed below Order (of Largest Connected Component) are computed over the largest connected component. * and ** respectively indicate statistical significance at the 95% and 99% level for the two-tailed test for the correlation coefficient with null hypothesis ρ=0