​

SOMETIMES TERRORISTS want credit for attacks, but at times, they prefer to wreak havoc without being blamed for the carnage. To prevent them from getting away with the latter and to aid law enforcement, researchers from Rice University in Houston, Texas, have developed a way to quickly deduce which terrorist group may be responsible for an attack. The tool was created by Derek Ruths, a Rice computer science Ph.D candidate, with the help of Sean Graham, a Rice undergraduate.

Ruths adapted spam filter technology to allow him to search a massive terrorism database to identify group attack patterns. A user plugs in details of an attack, and the tool names the groups with matching attack signatures.

The database Ruths uses was generated by the Institute for the Study of Violent Groups (ISVG), which is run by Richard H. Ward and Daniel Mabrey. ISVG’s administration and analytic development are at the University of New Haven in Connecticut, where Ward serves as a dean. Data collection is conducted at Sam Houston State University in Huntsville, Texas.

The ISVG database is made up of open-source information collected by researchers and includes 1,500 variables.

What intrigues Mabrey about Ruths’ tool is that it does a level of analysis on this rich source of data “that is too much for a single analyst to really think about critically,” he says.

“This can really be a kind of a starting point for analysts to start making this decision about who conducted an attack,” he notes.

Ruths coauthored a paper describing his research for Rice’s James A. Baker III Institute for Public Policy, along with Christopher Bronk, an institute fellow. The paper cited the recent terrorist attacks in Mumbai and stated that the computational tool, which uses the Python programming language, was able to pinpoint Lashkar-e-Tayyaba as one of the groups most likely behind the attack. Indian intelligence, as reported in the media, had identified that group early on as the operational source of the attack based on what they learned from the sole captured attacker. Ruths’ tool, once refined, could make a significant contribution in cases where no human intelligence is forthcoming.

“It’ll be valuable from the standpoint that, as soon as they start getting values back and properties and attributes of the attack, they can start plugging those in and seeing what groups come up,” Ruths says.

The tool has a long way to go before the intelligence or law enforcement community could consider using it, however. Mabrey says ISVG is continuing a relationship with Ruths on the tool. One issue is whether models could be developed that need fewer variables and work more quickly than the current version.

While the tool is intriguing, it should be only one among many, or “it could close minds or lead to group think,” says Jeremy Littlewood, director of the Canadian Centre of Intelligence and Security Studies at

Carleton University in Ottawa, Ontario, Canada.

Ruths agrees. “Ultimately the decisions…must be made by people who have the judgment to decide what’s right and wrong,” he says. “This is as biased to data as people are. If you give it wrong data, it will give you wrong results.”