GFD: A Weighted Heterogeneous Graph Embedding Based Approach for Fraud Detection in Mobile Advertising

Table 2

Classification results for all embedding models in AP (mean ± std).

Date	DeepWalk	Node2vec	Metapath2vec	WMP2vec
1st June	0.168 ± 0.082	0.343 ± 0.057	0.344 ± 0.065	0.384 ± 0.055
2nd June	0.318 ± 0.137	0.384 ± 0.079	0.342 ± 0.067	0.421 ± 0.084
3rd June	0.281 ± 0.075	0.439 ± 0.122	0.401 ± 0.110	0.459 ± 0.114
4th June	0.313 ± 0.073	0.335 ± 0.085	0.360 ± 0.096	0.409 ± 0.099
5th June	0.308 ± 0.121	0.379 ± 0.102	0.387 ± 0.074	0.411 ± 0.091
6th June	0.199 ± 0.113	0.230 ± 0.089	0.369 ± 0.097	0.404 ± 0.102
7th June	0.244 ± 0.054	0.297 ± 0.075	0.371 ± 0.094	0.353 ± 0.075