February 16, 2004
The Australian Financial Review
By Rachel Lebihan

The securities regulator is wielding a new weapon: an automated web spider designed to drive rogue financial advisers and investment swindlers from the internet's dark side.

The Scamseek system uses a search engine to crawl the net non-stop. During trials it has sparked a number of investigations into unlawful investment schemes and unregistered financial advisers.

It can also identify the notorious Nigerian letter scams, "phishing" - or fake email - rorts, and is being developed to recognise "share ramping" schemes that rely on false information on internet bulletin boards to help inflate company stock prices.

The director of the $2.2 million Scamseek project, the University of Sydney's Jon Patrick, said that though the Australian Securities and Investments Commission was the project's primary client, the software would also be of interest to other regulators, law enforcement agencies, and consumer organisations.

The project partners had already signed a deal with another group, but because of the sensitivity of the operation "it will probably never become public knowledge".

The Australian Taxation Office had also reviewed the Scamseek technology, which would help it to scan the internet for tax avoidance and evasion arrangements that may be promoted from offshore tax havens.

ASIC director of electronic enforcement Keith Inman said a final product was still some months away but the software already "exceeded expectations".

"It's giving us investigation leads far more efficiently than our manual processes ever did," Mr Inman said.

"We already have a couple of investigations afoot and have made half a dozen referrals to our counterparts in the United States."

In the past ASIC held "surf days" for 30 or more staff to manually trawl the internet and identify potential scams. On average, 1000 potential scam sites were picked out on these days, which led to 10 investigations and one warning or prosecution.

Mr Inman said there were huge efficiency gains in using Scamseek, which had a 75 per cent precision rate when detecting shonky investment schemes or financial advisers.

The software combines linguistic and statistical analysis of documents and extracts information about the organisers of the scam.

"It used to be that the chance of the regulator finding your one [scam] page amongst the 3 million added to the web every day meant the odds were stacked in your favour. Now that's not the case," Mr Inman said.

The software was developed by the Capital Markets Co-operative Research Centre, ASIC, the University of Sydney, Macquarie University, and SMARTS which designs fraud detection and analysis systems.