BY Jason Togyer - Fri, 2009-08-07 06:08
- B.S., computer science, Purdue University, 1982
- Ph.D., computer science, Carnegie Mellon University, 1989
With offices in 150 countries and 155,000 employees, PwC audits thousands of publicly traded companies, private corporations and non-profits. That makes reliable data-mining tools potentially one of the most important pieces of an auditor's workbench, says Steier, director of PwC's Center for Advanced Research in San Jose, Calif. "One big area we work on is fraud risk detection. There are millions and millions of data points --- there's no way that a manual review can find some of these things."
Tools developed by Steier's team look for anomalies --- for example, is a certain department paying prices inconsistent with industry norms? And they spot coincidences that are --- as a TV detective might say --- "a bit too convenient." (Do vendors share addresses or other relationships with employees?)
Ideas for some applications come directly from the field. "We deliberately look for high-risk, high-reward problems," Steier says. One internal social-networking application developed by the center can match any one of thousands of partners and staff with hard-to-solve problems by looking at resumes, past experience, reports they've authored and other data.
While much of their work is confidential, PwC researchers do publish and collaborate with peers, including those at Carnegie Mellon. PwC scientists recently worked with Christos Faloutsos, professor of computer science at SCS, and Mary McGlohon, a doctoral candidate in the Machine Learning Department, on applying algorithms that find patterns in social networks to looking for links in financial data.
Steier finds his work rewarding because of its broad reach. "The way that the financial statements of companies you invest in are audited are likely to be impacted by the work we do here," he says.
For More Information:
Jason Togyer | 412-268-8721 | jt3y [atsymbol] cs.cmu.edu