Search:  

 Blog     e-Newsletter       Resource Library      Directories      Webinars     Apps     EBooks
   AFCEA logo
 

Googlizing Intelligence

June 2010
By Maryann Lawlor, SIGNAL Magazine
E-mail About the Author

 

Spc. Adam G. Whitmire, USA, intelligence specialist, 101st Airborne Division, documents the serial numbers of cell phones after they were confiscated in Khost province, Afghanistan. Data from this type of find can become lost in the shuffle of thousands of intelligence reports. Software developed by Modus Operandi sifts through mounds of gathered intelligence information to deliver only pertinent data to analysts quickly.

Software digs through mounds of data to deliver gold nuggets to the front line.

Actionable knowledge will be available to commanders at lightning speed as the U.S. military and industry institute more adept methods to sift through terabytes of raw intelligence data. With the help of language-crunching software, intelligence analysts will be privy not only to crucial data about people, organizations, locations and weapons but also to the relationships among them. The key that unlocks the door to this obscure information is technology that enables computers to recognize and collate words and their meanings. In a matter of minutes, it then organizes the data in a way that would take weeks for a human analyst to accomplish.

Each of the services has found its own applications for this technology; however, they all agree that the solutions one company offers and continues to develop are unique in a number of ways. Modus Operandi Incorporated, Melbourne, Florida, has spent the last several years designing semantic software that tags text in a way that enables intelligence analysts to extract key information about designated entities. By employing the software, users can set the parameters for the data they seek and receive the equivalent of CliffsNotes that contain the information most relevant to their query. However, not only does the technology deliver documents containing specific words, it also culls relevant terms from a multitude of sources and places this information in the hands of analysts just as efficiently as Google delivers search results to the desktop.

According to Tod Hagan, director of intelligence, surveillance and reconnaissance software solutions, Modus Operandi, the company demonstrated the benefits of its solutions in a very simple and straightforward manner. A hundred intelligence reports were chosen directly from the field and given to intelligence specialists to analyze. At the same time, those reports were put through the company’s software. While the task took the intelligence analysts two weeks to complete, the Modus Operandi product identified 19,000 essential intelligence elements in only a few minutes, Hagan maintains.

The firm’s method uses natural language processing, which examines the structure of sentences and tags the essential elements of information, then relates these elements to each other and to a query. One feature that makes the company’s solutions stand out is their ability not only to pull out specific data as requested but also to recognize more general references to the same information. For example, the analyst may ask for documents that include a specific time. In response, the Modus Operandi software returns items that include the exact time as well as more general terms such as yesterday, today or tomorrow. In addition, the software is interoperable with all operating systems, which increases its usefulness.

Hagan points out that quickly citing data is not enough when it comes to identifying intelligence that literally could be the difference between life and death. The information that is delivered to the analyst also must have a high degree of accuracy. For example, the U.S. Air Force has been using a system that provides essentially the same solution as the Modus Operandi product with one big exception: precision. The service then decided to examine the benefits of the company’s products. Depending on the query, a 9 percent to 240 percent improvement in accuracy was achieved after integrating the company’s and service’s software.

Because the software was developed under a government contract, it has been and will continue to be designed with the government in mind. Military- and intelligence-specific terms are integral to the natural language-processing elements. The magic behind the curtain is its ability to accurately index words, assigning each a specific value.

Also because it was developed with federal funding, government organizations can obtain the software at no cost. Each of the services is taking advantage of this opportunity and is adding its own funds to create the intelligence-sifting package that best meets its needs.

For example, the U.S. Marine Corps currently is working with Modus Operandi on one of the company’s latest ideas: sharing the refined intelligence information via a semantic wiki. To Lt. Col. Scott E. Camden, USMC, deputy program manager for project management, intelligence, data diffusion and dissemination, Marine Corps Systems Command (MARCORSYSCOM), Quantico, Virginia, this capability will render immense benefits to warfighters. “We are using this technology to make the text reporting we have more valuable. We literally have Marine human reconnaissance teams risking their necks to generate intelligence reports that we currently have very little capability to fully analyze. Humans can only analyze so much data at a time,” Col. Camden explains.

 

Modus Operandi is not the only technology that keeps intelligence flowing. U.S. Air Force testers examine the new Valiant Angel system, which will provide additional intelligence details and increase warfighters’ situational awareness.

“This will help the analysts to find the silver needle in a stack of silver needles,” he states. “It is really about the work ratio for the analysts. Currently, analysts spend 70 percent of their time finding the data and 30 percent of their time analyzing the data. This [technology] has the potential to reverse these percentages.” In addition to identifying relationships and other valuable knowledge, this capability also identifies gaps in intelligence. “Now you know what you don’t know,” the colonel adds.

MARCORSYSCOM currently is working on a two-part effort. First, its goal is to be able to sift through intelligence data and find the links between intelligence reports, in particular the latest reports. Its second objective is to share those links in a wiki environment so that the data is easily available not only to Marines but also to the other services as well as military leaders. “This could have applications beyond Intellipedia,” Col. Camden states.

Currently, the tool for the Marine Corps is in prototype form and can be described best as “quiltware,” the colonel says. The technology readiness level, or TRL, needs to be matured, but Col. Camden says he does not believe it will take much to raise it to the level necessary for fielding. His estimate is that Marines will be able to use the capability around fiscal year 2011. Time to field is more about funding and human resources than it is about technical capability, he points out. Policy, doctrine and user manuals also must be developed before fielding, the colonel adds.

The capability is applicable to each of the services because they have similar problems and each uses some form of the Distributed Common Ground System (DCGS). The Marines actually became aware of the Modus Operandi solution through quarterly meetings with the U.S. Army, which already was working with the company. As a result, the Marine Corps jumped onboard the semantic convoy with the Army and provided additional funding to pursue capabilities specific to its own needs.

According to Larry Lashine, about two years ago the Army recognized the need to accelerate its investment in technologies that could close the gap between mounds of raw intelligence data and the analysts’ ability to scrutinize it. Lashine is the branch chief for DCGS-Army (DCGS-A) Systems, Software Engineering Directorate, at the Research, Development and Engineering Command (RDECOM), Communications-Electronics Command (CECOM), Fort Monmouth, New Jersey. One aspect of the challenge, Lashine says, is the need to create and maintain machine-processable vocabularies that enable automated systems to organize and exploit unstructured information.

This is a particular problem as unanticipated threats arise that result in new word clusters that a computer may not identify. In these cases, critical information is not tagged or organized in a way that analysts can use via computer-assisted means. “A simple example is when militants and terrorists began to use new types of improvised explosive devices [IEDs]. Terms such as IED, vehicle-borne IED [VBIED], animal-borne, boat-borne, even house-borne, are now used in intelligence reports to describe the enemy’s methods. Computer systems need to rapidly recognize these and other new terms so that attack patterns can be automatically detected,” Lashine explains.

In August 2009, CECOM awarded Modus Operandi a $70,000 contract to develop automated natural language sensor programs. Phase one of the contract, which RDECOM is sponsoring, is Vocabulary–Intelligence, Surveillance and Reconnaissance, or V-ISR. It centers on text-based data analytics for the Army’s ISR systems, including the DCGS. The company is developing “sensor bots” that monitor data feeds, extract vocabulary, relate it to intelligence analysts’ concepts of interest and then make the knowledge available to other software components.

Lashine points out that the information that is important to intelligence analysts is not only about people, places and events but also about what they mean. “Commanders need to know ‘why,’ ‘what does this mean for the future’ and ‘what are the enemies’ capabilities matched with their intentions.’ Only when analysts can bring together the disparate facts from multiple sources can they hope to help the commander understand his uncooperative and unwilling opponent’s intentions,” he notes.

The goal of Lashine’s work is to provide Col. Joseph Grebe, USA, project manager, DCGS-A, with a tool suite that the colonel can integrate into the DCGS-A and be used by all field commanders and soldiers in current operations. In addition, Lashine intends to share this capability with other programs of record such as the Aerial Common Sensor and several other DCGS organizations specific to services and the intelligence community. Furthermore, Lashine says he plans to offer agencies such as the U.S. Department of Homeland Security and U.S. Coast Guard the opportunity to benefit from the Army’s work. Finally, he believes that the technology could be incorporated into the Army’s Joint Unified Multi-Capable Protection System, or JUMPS, which provides users with access to advanced technologies that do not belong to a program of record.

The U.S. Navy has similar aspirations. According to a representative from the Submarine Combat System Program Office, the Modus Operandi approach offers a solution to the problem of information fusion and representation at the command level based on a service-oriented architecture. This is a capability that other vendors have not offered. Commanders will receive data from multiple sensors and sources that are combined to reduce the complexity of the tactical operational picture and enable faster decision making, the representative says.

A proof-of-concept demonstration of the capability has been underway since April and is scheduled to wrap up this month. A second demonstration is planned for 2011 in the April to June time frame. After the technology successfully demonstrates its capability and value, it would be made available for transition into a future combat system using an advanced development four-step process.

In March, the Air Force Research Laboratory awarded a $425,000 project to Modus Operandi to expand the service’s intelligence processing and analysis capabilities. The software will significantly improve intelligence analysis of the existing application called Web-based Threat HUMINT [human intelligence] Reporting Evaluation Analysis and Display System (WebTHREADS). The company will leverage capabilities the Air Force has developed and deploy its Wave Exploitation Framework, which will enhance WebTHREADS’ ability to filter, parse and persistently sort through large amounts of data.

WEB RESOURCES
Modus Operandi Incorporated: www.modusoperandi.com
U.S. Marine Corps Systems Command: www.marcorsyscom.usmc.mil
U.S. Army Research, Development and Engineering Command: www.cerdec.army.mil
U.S. Navy Submarine Combat System Program Office: https://acquisition.navy.mil/rda/home/organizations/peos_drpms/peo_subs
U.S. Air Force Research Laboratory: www.wpafb.af.mil/AFRL

 

Comments

Saw this article in my June AFCEA SIGNAL magazine. It relates to the Search and Metadata work for NEP.

By Norm Heath