Short bio: Computer Scientist, FOSS supporter (read more)
Tux Machines (TM)-specific
The National Security Agency is teasing intelligence information out of massive federal databases via a program that uses innovative methods to discover hidden results and improve analyst performance.
The Glass Box program has been using unclassified, open-source information for three years to produce information used by the Novel Intelligence from Massive Data program, according to contractor Battelle Memorial Institute of Columbus, Ohio.
NSA recently awarded Battelle a $1.507-million contract to continue its Glass Box work from last January through June 2006. The agency had sought sources for Glass Box work in September 2004, according to Federal Sources Inc., a consulting firm that tracks IT contracts.
According to FSI, the government did not receive any responses to its sources-sought notice and decided last November to negotiate with a list of previously identified potential contractors. Eventually, only Battelle submitted a proposal. The vendor and the government then negotiated an agreement.
According to contract documents, Glass Box is designed partly to evaluate intelligence algorithms developed by other, presumably classified, NIMD activities.
The program also is intended to develop the Glass Box Analytic Environment, a hub task of the NIMD project. The contract documents describe NIMD as a research and development project intended to last for several years to discover the "novel intelligence" that lurks undetected in the intelligence community's databases.
Novel intelligence falls in the category of information that causes intelligence analysts to gain new understanding of previously unappreciated or misunderstood threats, according to the contract documents.
"This information may emerge from discoveries in the content and patterning of new data, or discoveries resulting from 'out-of-the-box' analytical reinterpretations of existing data," according to the government. "Novel intelligence often equates to 'suspicious revelations' in one-time or first-time events, or in hidden connections between seemingly independent events or entities, or in puzzle pieces collected long ago but that suddenly make sense in light of recent observations."
Battelle responded to an e-mail inquiry by stating that analysts working on the Glass Box experiment have processed more than 70 intelligence analysis tasks and produced more than 80 reports. "More than 48 gigabytes of data are now contained in the Glass Box database and file store," Battelle said.
"From examining the Glass Box data, researchers have created visualizations of the analytic process, identified inefficiencies in the current analytic work environment, improved search techniques, created user models that help tools understand and predict analyst behaviors, and more," according to the research institute.