CBC: Counterterrorism Tools Take Aim at Deceptive Emails.

[Tandalayo_Scheisskopf]

KINGSTON, ONT. - A new approach that tracks word patterns in e-mails has the potential to help nab terrorists, say computer scientists at Queen's University.

Every day, counterterrorism experts use a top-secret electronic intelligence gathering network called Echelon to intercept hundreds of thousands of private e-mails, faxes and phone calls, including in Canada. Traditional intelligence surveillance programs search for a list of key words. But the large volume of e-mails and unsophisticated approach of searching for key words that terrorists likely know to avoid has hampered counterterrorism efforts. Computer scientists have long theorized that analysing word patterns in e-mails could reveal key figures in a group communicating over the net, without having to read each message.

At Queen's University in Kingston, Ont., David Skillicorn and his colleagues have applied this technique to tracking e-mails. "In 500,000 e-mails, you can't do anything without some kind of automated tool," Skillicorn said. "The trick is to pick out the top 1,000, even 100, you might want to look at it more detail." Skillicorn's group honed in on a common feature of deceptive behaviour: people leaving a "signature" through their carefully chosen words when they practise to deceive. They've developed a surveillance program to detect deceptive word patterns in suspicious e-mails. The team was able to test their approach after federal energy regulators in the U.S. unexpectedly posted 1.5 million e-mails from employees of the disgraced American energy conglomerate, Enron, on the web.

Since the Enron e-mails were sent before, during and after the accounting crisis, Skillicorn's team was able to track changes in the correspondence. As the crisis deepened, the e-mails contained fewer personal pronouns, more negatives and simpler sentences as writers attempted to distance themselves from what they were saying, the researchers found. "Mostly it's about looking for word patterns and seeing which ones are the most unusual," the computer scientist said.

http://ottawa.cbc.ca/regional/servlet/View?filename=ot-emails20050615

[consuming]

1. Why?