Intelligent Risk Control Practices and Architecture by Shumei Technology
This article presents Shumei Technology's comprehensive approach to fraud prevention, detailing the scale of black‑market losses, typical abuse scenarios, challenges of traditional defenses, and the design of a full‑stack, AI‑driven risk control system that combines device, behavior, and content detection with real‑time, multi‑cluster deployment and case studies from banking and live‑stream platforms.
The presentation introduces the background of fraud in the internet ecosystem, highlighting massive financial losses caused by black‑market activities and the difficulty of defending against sophisticated attacks across UGC platforms, e‑commerce, and financial services.
It outlines the challenges of existing risk control solutions: weak defensive capabilities, poor timeliness, and slow evolution, emphasizing the need for a more robust, adaptive system.
Shumei offers two core products—Tianwang for device and behavior fraud detection, and Tianjing for UGC content moderation—forming a full‑stack risk control framework that includes four subsystems: deployment, strategy, portrait, and operation.
The deployment architecture is described in detail, covering the main interface, business logic layer, core service layer with an intelligent decision engine, strategy tuning layer, and foundational services such as blacklist management, all built on a micro‑service model with high availability across multiple data centers.
Key technical components include a Rete‑based rule engine for fast strategy execution, a hybrid model stack (expert system, unsupervised clustering, supervised learning) for device, behavior, and group detection, and AI models for text, image, audio, and facial recognition.
Real‑time risk control is achieved through a low‑latency data flow, device SDK integration, and a workflow engine that orchestrates preprocessing, model scoring, and post‑processing, ensuring sub‑50 ms response times.
Case studies demonstrate the system's effectiveness: a bank marketing anti‑fraud deployment reduced monthly losses by approximately ¥25 million with a 99 % interception rate, and a major live‑stream platform achieved over 99 % detection of fraudulent accounts and advertising content.
The article concludes with a Q&A on the role of the testing engine and feature definition, reinforcing the importance of continuous model and strategy optimization in combating evolving fraud threats.
DataFunSummit
Official account of the DataFun community, dedicated to sharing big data and AI industry summit news and speaker talks, with regular downloadable resource packs.
How this landed with the community
Was this worth your time?
0 Comments
Thoughtful readers leave field notes, pushback, and hard-won operational detail here.
