Graph Computing at Ant Group: From Fraud Prevention to Industry‑Wide Impact

Graph computing, a data structure that models relationships between objects, has become a core technology for Ant Group’s anti‑fraud system, enabling real‑time risk identification within milliseconds of a transaction attempt.

Since 2015 Ant has built its own graph database, GeaBase 1.0, which powered the Alipay network during the 2016 Double‑11 shopping festival and later scaled to billions of relationships, becoming indispensable for the platform.

Key team members such as Chixiao and Qiao Qi, both PhDs from Tsinghua with experience at Microsoft and startups, lead the Ant graph‑computing team. Their work includes the large‑scale graph platform TuGraph, which supports real‑time and temporal graph analysis, achieving millisecond‑level latency for fraud and money‑laundering detection.

Beyond anti‑fraud, the team contributes to industry standards (e.g., ISO GQL) and pushes innovations like perfect‑hash‑based real‑time graph storage, which dramatically improves query performance and is slated for production deployment.

The article emphasizes the importance of translating research into products, noting that while Ant’s technology is mature, publishing academic papers remains limited; the team aims to bridge this gap and expand graph computing into domains such as epidemic modeling, public safety, and broader industrial applications.