The approach taken by eBay turns usual ideas about automated fraud detection on its head. Rather than focusing on the changing patterns employed by bad actors to circumvent protective barriers, the proposition instead analyses instances of good behaviour.

In a paper published on preprint server Arxiv.org, San Jose-based Utkarsh Porwal and Smruthi Mukund note that patterns of good behaviour do not change with time and the data points that represent this form of conduct have consistent spatial arrangements. Porwal and Mukund suggested a clustering method for identifying outliers and to later formulate a score, which would determine consistency and in turn, good behaviour.

Using a public dataset of credit card transactions made in September 2013 by European cardholders over two days from Google’s data science community platform Kaggle, they analysed 284,807 samples, of which 492 were fraudulent. Data revealed that precision was high until a recall of 0.4 and then dropped, suggesting that 40% of the fraud cases were identified with high precision.

“Often the challenge associated with tasks like fraud and spam detection is the lack of all likely patterns needed to train suitable supervised learning models. This problem accentuates when the fraudulent patterns are not only scarce, they also change over time. Change in fraudulent pattern is because fraudsters continue to innovate novel ways to circumvent measures put in place to prevent fraud. Limited data and continuously changing patterns makes learning significantly difficult. We hypothesize that good behavior does not change with time and data points representing good behavior have consistent spatial signature under different groupings,” the report states.