Answers are grounded in this paper and cite their sources.
Shows novel B2B payment fraud detection framework, suggesting improved methods are needed.
Academic fraud-detection research is dominated by business-to-consumer (B2C) credit-card data, and there is no openly available benchmark for business-to-business (B2B) payment fraud, whose signals are inherently relational rather than transactional. We release SynB2B-Fraud, a fully synthetic, openly licensed (CC BY 4.0), DOI-citable dataset of 100,000 buyer→supplier invoice transactions spanning eighteen months, in which five B2B fraud topologies—vendor injection, invoice cycling, payment-term manipulation, shell-supplier rings, and wire redirection (business email compromise)—are injected using explicit, timestamp-respecting firing rules. Every fraudulent row carries a binary label and its specific topology. The dataset ships with a seed-deterministic generator. We report reference tabular baselines and a recommended evaluation protocol, and show that raw-feature classifiers capture only part of the signal—motivating relational and graph-based methods such as STREAM-BSG.
No takes yet. Share an insight, caveat, or question.
Abhishek Sharma (2026) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: