iBGP: A Bipartite Graph Propagation Approach for Mobile Advertising Fraud Detection

Table 2

iBGP continuously maintains the highest Kappa value, despite various levels of camouflage or size of fraud users.

Graphs	iBGP	NodeProp	EdgeProp	Hits-o	BP-a
SG + 0% R	1	0.9620	0.9996	0.9624	0.7667
SG + 20% R	0.9598	0.9298	0.9489	0.9305	0.7469
SG + 40% R	0.9342	0.8976	0.9153	0.9002	0.6875
SG + 60% R	0.8894	0.8402	0.8386	0.8390	0.5369
SG + 80% R	0.8366	0.7432	0.8060	0.7775	0.4825
SG + 40% R^-	0.9327	0.8991	0.9252	0.9021	0.1833