Recently, DataVisor CEO and Co-Founder Yinglian Xie had the opportunity to speak at Strata+Hadoop World in San Jose and discuss how we leverage Spark to catch even the most advanced fraudsters lurking in today’s online services. One of the main issues with today’s consumer-facing websites and mobile apps is that they are measured by the size and growth of their user account base. These users act as both contributors of content and a channel for monetization and with so much riding on them, they can also be an Achilles heel. Well-organized crime rings create millions of fake user accounts to hide among billions of legitimate users. These large groups of bad guys have become adept at mimicking the behavior of real users, making them difficult, almost impossible, to discern with traditional rules-based security solutions.
In her talk, Yinglian uses real-life case studies to describe the anatomy of modern attacks and how these fraudsters are able to easily skirt traditional security solutions. She demonstrates how the subtle signals from these attack campaigns can be detected, even in their early incubation “sleeper cell” phase, using unsupervised algorithms running within a massively scalable Spark framework.
If you’re interested in taking a look at her talk, or any of our other resources, you can find them here.