Method and apparatus for establishing an abnormal account detection model

Abstract

An anomalous account detection model building device includes a storage module and a processing module. The storage module stores multiple sets of account information corresponding to multiple accounts, and an expert rule system for determining whether account information is anomalous. Each set of account information includes multiple transaction records. The processing module uses the expert rule system to determine whether the account information is anomalous based on the transaction records of the account information, dividing the account information into a rule-abnormal group and a rule-non-abnormal group. Based on all account information in the rule-abnormal group, a machine learning algorithm is used to build an artificial intelligence model for determining whether account information is anomalous, and then an anomalous account detection model including the expert rule system and the artificial intelligence model is generated.

Images