Anti-fraud needs to be based on high-quality data and use methods such as association, classification, clustering, and anomaly mining to build multi-layer, multi-dimensional, and multi-structure anti-fraud and quantitative risk control models. The anti-fraud data sources of financial institutions include identity information, credit information, social information, consumption information, behavioral information, etc. Through the portrait characteristics of tens of millions of people labeled as C-end users (whether for loans, credit card applications, or other financial activities), more accurate identification of anti-fraud behaviors can be achieved. General data sources, in addition to own data, also include external data cooperation. For example, MobTech's company produces data on mobile devices. As we all know, Chinese Internet users are mainly concentrated on the mobile Internet.