Browse the Archive

KSP (aka the 'BAG'): Connecting the Dots

Collapse Details

DYNAMIC PAGE -- HIGHEST POSSIBLE CLASSIFICATION IS TOP SECRET // SI / TK // REL TO USA AUS CAN GBR NZL (U//FOUO) KSP (aka the 'BAG'): Connecting the Dots FROM: SIGINT Communications Unknown Run Date: 09/03/2003 (S//SI) Hundreds of databases, trillions of pieces of information... how can a SID analyst see patterns and extract what is needed? This is the question that the Knowledge System Prototype (KSP, which includes the "Big Awesome Graph" or "BAG") is striving to answer. While other programs such as Trailblazer and Rebuilding Analysis are working on unifying SID's databases and computer applications, what KSP aims to do is a little different. It wants to help find pieces of information scattered throughout our hundreds of databases and show how they interrelate with each other. KSP will enable analysts to "connect the dots" across vast amounts of data - a process that would often take forever for an analyst to do by hand. (S//SI) It should be noted that KSP is not just an idea or a way of doing things - it's actually going to be a real "box" that does something. Namely, it will help analysts sort through information from various sources - reference information, target knowledge, contact info, raw data - to see trends that might not otherwise be apparent. (S//SI) Let's look at a simple example to see how it would work. First, an analyst may ask KSP to find all organizations in several named countries that have the word "trading" in them. Once this information is displayed, the analyst may then have those organizations' known email addresses and phone numbers added to the display. From this, certain trends may appear. For example, perhaps two of the organizations are companies that share a common phone number an indication that the same company may go by multiple names. Or perhaps two companies have similar email addresses, even though they are located in different countries - again, a sign that they may be affiliated. (S//SI) One interesting feature of KSP is that it will independently search through the available data to find and store additional occurrences of this scenario. So, in this example, it would find all other cases where companies with different names use the same phone number or email address. Since KSP is "anticipating" questions, it can have the answers already prepared and ready to go for the next time an analyst asks that question. As a result, the analyst will not need to wait for KSP to crunch through trillions of records to answer every query; often, it will be able to provide an answer very quickly. (S//SI) Using the current prototype, the KSP team expects to have a demonstration capability prepared by November 2003. Then they plan to work with users to refine it, and incrementally deploy it out to the workforce. Watch for further developments! Check out the Knowledge Enablement web page for further details. "(U//FOUO) SIDtoday articles may not be republished or reposted outside NSANet without the consent of S0121 (DL sid comms)." DYNAMIC PAGE -- HIGHEST POSSIBLE CLASSIFICATION IS TOP SECRET // SI / TK // REL TO USA AUS CAN GBR NZL DERIVED FROM: NSA/CSSM 1-52, DATED 08 JAN 2007 DECLASSIFY ON: 20320108

Filters SVG