Social scientists and mathematicians join the hunt for terrorists.

• Author: Beidel, Eric
• Position: HOMELAND DEFENSE

During a recent visit to Washington D.C., New York City Police Commissioner Raymond Kelly told insiders and media that future terrorist attacks in the United States will resemble the failed car bombing in Times Square. While lacking the ferocity of 9/11, these small-scale attempts will be carried out by U.S. residents who are radicalized over the Internet, he said.

[ILLUSTRATION OMITTED]

The Senate Committee on Homeland Security and Governmental Affairs looked at how the Internet has propelled violent Islamic extremists and the homegrown terrorist threat. The committee found that defeating domestic terrorism required efforts beyond classified intelligence and law enforcement programs.

Social scientists, computer whizzes and mathematicians have offered to help. Since 9/11 the academic community has produced countless theories, methods and formulas in an attempt to help federal agents combat terrorism. Some of these professors will hit on something, and the rest will end up in blind alleys, they admit.

"At the end of the day, finding terrorists is not an easy job," said V.S. Subrahmanian, a computer science professor at the University of Maryland who invented an algorithm to find patterns in social networks.

As networking sites like Facebook grow in popularity and make it easier to access information about Internet users, new methods for rummaging through data and connecting dots will emerge, Subrahmanian said. His team has come up with COSI, short for cloud oriented subgraph identification, a computer program that within seconds can find complex connections among billions of links in a social network.

Beyond simply finding another person interested in basketball, COSI could locate someone who not only likes basketball but also shares three mutual friends, plays video games, reads Ernest Hemingway and speaks Spanish. In a research paper, Sub-rahmanian's team provides examples of these "complex queries." One scenario takes a stab at a financial network involving a shady bank and pairs of alleged criminals: "Find all vertices (v1, v2) such that v1 wired money to Bank1 and Bank1 received a wire from v2 and both v1 and v2 have a common friend v3 who has been labeled suspicious."

For average Facebook users, this means that search engines could better match them with friends and suggest groups with more closely aligned interests and concerns. For terrorists communicating in a secret network, it means having to find trickier ways to hide. Dutch researchers previously developed mathematical models to show what the structure of a covert network looks like--the proverbial needle. Now, COSI...