Anti-fraud model generation method, apparatus, apparatus, and storage medium

Abstract

The invention discloses a generating method, an apparatus, a device and a storage medium of an anti-fraud model based on machine learning. The method comprises the following steps: obtaining preset fraud rule flow information and sample information of a user transaction; putting the sample information into the preset fraud rule flow information for analysis to obtain hit branch information and similar branch information; subdividing the hit branch information and the similar branch information to generate the target branch information; extracting the characteristic information of fraud behavior from the target branch information and training the preset anti-fraud model to generate the target anti-fraud model. By acquiring hit branch information and similar branch information in fraud ruleflow information, as fraud characteristic information, fraud characteristic information is trained to the preset anti-fraud model, so that the effective characteristic information in the fraud rule flow is extracted in advance, and the anti-fraud model is trained through the effective characteristic information, so as to reduce the model construction time and cost.

Description

technical field [0001] The present invention relates to the technical field of data processing, in particular to a method, device, equipment and storage medium for generating an anti-fraud model. Background technique [0002] In general, the modeling of anti-fraud rules usually requires the following process. First, the user's transaction data is preprocessed. For the transaction data that is seriously unbalanced, under-sampling is also required. In addition, in order to realize the transaction data For processing, it is also necessary to convert the transaction data into a data format supported by the model, aggregate the features extracted from the transaction data into vectors, divide the transaction data into a training set and a test set through the vectors, construct an anti-fraud model through the training set, and use the test set to conduct It is predicted that this modeling process is complex and requires a lot of time and labor costs. Contents of the invention ...

AI Technical Summary

Problems solved by technology

[0002] In general, the modeling of anti-fraud rules usually requires the following process. First, the user's transaction data is preprocessed. For the transaction data that is seriously unbalanced, under-sampling is also required. In addition, in order to realize the transaction data For processing, it is also necessary to convert the transaction data into a data format supported by the model, aggregate the features extracted from the transaction data into vectors, divide the transaction data into a training set and a test set through the vectors, construct an anti-fraud model through the training set, and use the test set to conduct Forecast, this modeling process is complex and requires a lot of time and labor costs

Images