基于银行账户亲密度网络推理的团伙预测研究

In recent years, research on the law of capital transactions in the case of stakeholder-type illegal financial activities has attracted the attention of researchers. In order to solve the problem of using the bank transaction data to actively discover the abnormal account criminal crime partners, a crime partner prediction method based on the asymmetric intimacy network of the bank accounts was proposed. Firstly, a general network model for bank account transactions was established to embed time-series transaction data into the network structure. Then, using the direct and indirect transaction relationship information of the node, an account asymmetric intimacy calculation method was proposed. Finally, using the asymmetric interaction information of the nodes on the intimate network, the abnormal tendency index of the nodes is obtained. The experimental results on the actual data of the multi-level marketing group show that the crime partner prediction method based on the intimacy network can effectively find potential pyramid shelling members.