UMD Tool Predicts Leadership of Terrorist Networks
COLLEGE PARK, Md. – The loss of a terrorist or criminal network's leader—whether through imprisonment, change of allegiance or death—can create a vacuum in which subordinates jockey for position or splinter into factions.
Rather than wait to see how these scenarios play out, U.S. intelligence analysts could soon have a new tool to help predict who might rise to the top of a terrorist or criminal network, and whether the redefined organization has an increased ability to carry out its activities.
A University of Maryland research team developed this analytics tool, known as STONE (Shaping Terrorist Organizational Network Efficacy), “to minimize the impact of these organizations,” says V.S. Subrahmanian, a professor of computer science who is leading the UMD effort.
The UMD team has used open-source data to hypothetically test the software platform on four known terrorist organizations: al-Qaeda, Hamas, Hezbollah and Lashkar-e-Taiba, perpetrators of the November 2008 attack on Mumbai, India.
STONE was able to predict with 80 percent accuracy what individual would rise to take on a leadership role when a terrorist leader was removed, Subrahmanian says. The data the Maryland team used was unclassified, and included information such as how long a person was actively involved with an organization, the specific role they had, and the roles of others they were directly associated with.
U.S. government analysts and decision-makers with access to a “more complete” picture of these organizations can input their own data into STONE, increasing the tool's accuracy, Subrahmanian says.
“This is a not a computing tool that tells [analysts] what to do,” he says. “It is something that can help them better understand the situation or situations they are dealing with, which can ultimately decrease the efficacy of these organizations.”
Law Enforcement and Business Applications
Subrahmanian says STONE could also potentially be used to evaluate leadership changes in criminal networks and in business, for example identifying who will replace a corporate CEO or who will step into a new role in a drug network. However, he and his colleagues have not yet tested it for these applications.
The Maryland researchers—Subrahmanian, Francesca Spezzano and Aaron Mannes, all associated with the University of Maryland Institute for Advanced Computer Studies—will present a paper on their work at an international conference on Aug. 27. The Advances in Social Networks Analysis and Mining conference in Niagara Falls, Canada, is sponsored by the Institute of Electronics and Electronics Engineers and the Association for Computing Machinery.
