A new Internet document classifications system that employs linguistics theory to identify online breaches of the Corporations Act is set to become the latest weapon in the Australian Securities and Investments Commission's (ASIC) arsenal against Australian Web scammers.
Dubbed "Scamseek", the new technology claims to have the potential to assess and aggregate the risk associated with information on a Web site using a combination of language analysis and Web scanning technology.
Scamseek will be developed by ASIC through a new joint research initiative with the Capital Markets Cooperative Research Centre (CMCRC), Security Markets Automated Research Training and Surveillance (SMARTS), Sydney University and Macquarie University. The project has an overall budget of more than $1 million.
Team leader for the CMCRC and the University of Sydney professor, Jon Patrick, said the system would use new theories on textual meanings "to unravel the deep linguistic features that will enable us to detect scam proposals no matter what surface form of language they use."
As well as utilising language technology, the project will also apply a specialist Web spider to search out potential Web sites, using technology developed by CMCRC member Security Markets Automated Research Training and Surveillance (SMARTS).
ASIC director of electronic enforcement, Keith Inman, said that while the linguistics technology behind Scamseek was new, ASIC had been conducting scans for potentially suspicious Web sites for several years.
Initially, scans were carried out through routine "surf days". These involved scouting the Internet for potentially suspicious sites using ASIC's specially grafted list of scam keywords. Although the theory behind the surf days was sound, Inman said the team's success rate was low: for every one or two suspicious sites found, 1000 legitimate sites were also analysed.
In a bid to hone down the ratio of legitimate versus suspicious sites, the team developed an automatic tool that scanned sites using ASIC's keywords list. The process, however, was still too "labour intensive", Inman said.
Unlike ASIC's current methods, the Scamseek tool would elaborate on the keywords list by devising a "style sheet", analysing not just the words used on an offer but the nature and relationship of the terms throughout the offer itself, he said.
Because the Scamseek tool will be automatic, it will allow ASIC to scan the Internet 24 hours a day.
"The system we are developing through this joint project will be an eye that never sleeps, constantly seeking out sites that we can take action against," Inman said.
The first phase of the research project is expected to take six months. If fruitful, the CMCRC will launch a commercial version of the tool.
Inman said that based on the preliminary work already done by the University of Sydney's linguistic department, ASIC was confident the tool would become an integral part of their crack down on Internet scams.
Although Scamseek will help to identify potential fraudsters, Inman also urged people looking for more information on ASIC or wanting to lodge a complaint to visit ASIC's website: http://www.asic.gov.au
