论文阅读笔记:Graphconsis---Alleviating the Inconsistency Problem of Applying Graph Neural Network to Fraud

Alleviating the Inconsistency Problem of Applying Graph Neural Network to Fraud Detection

论文阅读笔记:Graphconsis---Alleviating the Inconsistency Problem of Applying Graph Neural Network to Fraud

  • context inconsistency:context的意思是上下文;语境;环境。inconsistency的意思是不一致性。节点 v1v_1 的context inconsistency指,节点v1v_1的邻居里,有fraudster,也有beign user.
  • Feature inconsistency: 节点v1v_1是 fraudster 节点,它的邻居里,节点5是 fraudster 节点,节点4、6、7都是benign user,benigin user和 fraudster 节点的features 是不一样的,所以在用图神经网络聚集邻居信息时候,不能单纯的聚集邻居信息。
  • Relation inconsistency:上图(left)中有两种关系,relation I (虚线表示)和 relation II(实线表示)。在 relation I 下,fraudster v1v_1连接了fraudster v5v_5,还连接了benign user v4v_4v6v_6v7v_7;在 relation II 下,fraudster v1v_1连接了fraudster v2v_2v3v_3。relation I 和relation II 是不一样的,所以在处理的时候,我们需要采用不同手段,这就叫关系不一致性。