channel-separation-makes-it-hard-for-organizations-to-catch-bad-guys-without-entity-resolution-to-put-the-pieces-together

When it comes to fraud and other criminal acts, many organizations are barely catching the idiots, let alone the really bad guys. The idiots are not so important, since they’re likely to run out of gas on the way to the operation or take a wet match to the fuse anyway. On the other hand, the clever bad guys are always a big problem and, in most cases, they go completely undetected.

Based on a few decades of building bad guy hunting systems ranging from Las Vegas security and money laundering to insider threat and fake identities — I’ve learned there is one primary deception tradecraft that holds true for all clever bad guys, across all domains.

I like to use the words “Channel Separation” to describe this.

No clever bad actor uses the same name, address, date of birth and email address across all of their transactions. Instead, they intentionally obfuscate their identities to prevent others from making sense of what’s happening.

Noodle this:

channel-separation-the-primary-deception-tradecraft-blog-jeff-jonas

You and I use Channel Separation too by the way. If you have ever sent someone an encrypted document and then called to share the password, this is also a form of Channel Separation.

If you hope to find such criminals, you must be able to perform Channel Consolidation:

The magic that makes Channel Consolidation possible is entity resolution. But not just any kind of entity resolution. This amazing feat of strength requires an ability called “entity-centric learning” that isn’t found in traditional record matching methods.

To learn more about entity-centric learning watch this video called Entity Resolution 2.0 in Slow Motion while paying particular attention to record 10.

Senzing has created a special kind of AI for entity resolution that performs Channel Consolidation with exceptionally low false positives, without humans tuning and training the system. We do it in real time over very big data.