Project background

The business of a licensed consumer finance company is growing rapidly. The traditional manual credit approval process cannot support an average daily application volume of 10,000, and the risk control model is insufficient to identify new fraud methods.

Solution

• Multi-source data fusion: Integrate multi-dimensional data sources such as credit reporting, operators, e-commerce, social networking, etc.
• Graph neural network anti-fraud: Identify gang fraud and associated risks based on GNN
• Intelligent Approval Decision Engine: Automates the entire credit approval process, supporting second-level decision-making

Project Results

The approval time is shortened from 48 hours to 3 minutes, the defective rate is reduced by 1.2 percentage points, and the average daily number of applications processed increases to 50,000.