Graph Intelligence LayerKaggle IEEE-CIS Entity Linkage1.54x Fraud Lift Ratio

Entity Linkage & Multi-Account Fraud Ring Detection

Evaluates shared hardware devices (`DeviceInfo`, OS, Browser), card fingerprint clusters (`card1`–`card6`), and localized network regions (`addr1`/`addr2`) to uncover sleeper accomplice nodes via PageRank risk diffusion.

4 Rings Detected

0 Linked Nodes

Empirical Verification: Kaggle IEEE-CIS Real Entity Linkage Dataset

Validated on Kaggle IEEE-CIS Real Entity Linkage Dataset (Hardware, Card & Network Fingerprints). Ingests genuine hardware device telemetry (`DeviceInfo`, `id_30`, `id_31`, `id_33`), composite card hashes (`card1`–`card6`), and billing location coordinates across 75,000 transactions. Identified 1,647 multi-account clusters and 97 high-density fraud syndicates with an empirical 1.54x fraud risk lift over the global random baseline.

Entity Graph Canvas

Flagged Fraud Graph-Elevated Accomplice

Loading interactive entity graph canvas...

Scroll to zoom · Drag to pan · Click node to inspect risk propagationForce Layout: Barnes-Hut