MUMBAI, India, Sept. 22 -- Intellectual Property India has published a patent application (202611110421 A) filed by Srinivas Neelam on September 15, 2026, for Cloud-Based Cross-Account Transaction Graph System For Real-Time Financial Fraud Detection And Adaptive Risk Propagation.
Inventor includes Srinivas Neelam.
The application for the patent was published on September 18, 2026, under issue no. 38/2026.
Abstract: A cloud-based financial fraud detection system is disclosed for identifying coordinated fraud across multiple financial accounts in real time. The system receives transaction events and maintains a heterogeneous temporal cross-account transaction graph containing account nodes, transaction-associated entity nodes, and time- associated relationship edges. For a graph-changing event, a seed risk value and an event-specific propagation budget are generated, and a bounded affected subgraph is selected according to expected marginal risk, temporal relevance, and processing constraints. Provenance-bearing signed risk messages are propagated through the bounded subgraph using relationship-specific transfer weights, temporal decay, confidence, saturation, cycle damping, path deduplication, and counter-evidence. Direct risk and propagated risk are maintained separately, with propagated contribution being selectively capped unless independently supported. An adaptive graph partition controller can migrate or replicate hot graph regions according to cross-partition communication and decision latency. A fraud decision engine generates a machine action and a pseudonymous explanation path for a transaction-processing system, thereby reducing unnecessary full-graph computation while enabling low-latency detection of coordinated cross-account fraud. Accompanied Drawing [FIGS. 1-2].
Disclaimer: Curated by HT Syndication.
