Amazon Filters for Insurgent-Hunting

Amazon recommends books based on collaborative filters -- algorithms that examine how other people shop, and make suggestions accordingly. The Air Force wants to use the same technique, to help intelligence analysts find terrorists and insurgents. The service has handed out a contract to Utica, New York's Black River Systems to get the system started.

The goal of the program is to "appl[y] proven commercial collaborative learning technologies to the challenge of adapting and defeating the ever-changing tactics of today's adversaries."

Collaborative techniques, such as collaborative filtering, have been successfully implemented in dynamic commercial environments with impressive results. For example, Amazon.com has implemented its own unique collaborative filtering algorithm which is responsible for an amazing 35% of its sales as a result of recommendations from other customers with similar preferences and experiences... [I]t will [soon] be possible to continuously communicate in real-time with a population of analysts that can provide valuable and timely feedback to allow an automated system to adapt itself to a constantly changing environment This effort will apply collaborative techniques to... automatically detect and recognize both the known and evolving patterns of the dynamic adversary.

But this isn't the only "cognitive fusion" project that the Air Force is backing. Soar Technology, out of Ann Arbor, Michigan, is adapting its "intelligent, reasoning data pipeline" to intelligence work. "Our approach has been to develop a framework of cooperative intelligent agents that model the warfighter, the warfighter’s tasks, the common operating picture, and agent-generated inferences about the incoming
data streams and display systems to customize data delivery." Now, the company wants to add "the analysts's domain knowledge and cognitive power" to the mix.