Who Is Holding the Card? Machine Learning for Battling Friendly Fraud/First-Party Misuse
Defeating fraud begins with understanding fraudulent behaviors, in order to correctly identify who the fraudulent agent is behind a transaction. The differences between first and third parties constitute the foundation of fraud knowledge and this distinction is a key component in mitigating the problem in an efficient way.
Machine Learning (ML) is widely applied in the field of fraud detection, but is not often used as a way to approach post-payment fraudulent behaviors, especially for identifying what is first-party misuse as opposed to third-party fraud.
This presentation explores the use of ML to better scale what has traditionally been a lengthy manual process of reviewing chargebacks. This contributes to the recovery process by giving insights for representing chargebacks and training adaptive detection ML models in a more efficient way.