Online retail, eCommerce, frequently falls victim to fraud conducted by malicious customers (fraudsters) who obtain goods or services through deception. Fraud results in a significant monetary damage to online retailers. This damage can be mitigated by cancelling fraudulent orders in a timely manner. However, large online retailers receive 100,000s of orders per day on which manual analysis and manual cancellation cannot be applied. In this project we aim to develop automated techniques for (1) detecting fraudulent orders and (2) helping to prioritize the manual analysis of orders most likely to be fraudulent. We particularly focus on the identification of organized fraud in which groups of professional fraudsters place several fraudulent orders to maximize their gain. This project is realized in collaboration with, and funded by, Zalando Payments GmbH, which handles payment for the largest online apparel retailer in Europe: Zalando.

Conference paper publications

Demos & Posters

  • Secure Systems Demo Day 2019: Sebastian Szyller, Samuel Marchal: Detecting E-commerce Fraud with Large Scale Categorical Clustering, (May 29, Aalto University, Finland), poster

Source code

Recursive Agglomerative Clustering (RecAgglo) for categorical data (