Does Data Mining Work?

The reason I ask is because Homeland Security has closed down another data mining project. Is there anybody out there who can help these guys?Earlier this week, the Department of Homeland Security closed down its Analysis, Dissemination, Visualization, Insight and Semantic Enhancement program, known as ADVISE. The data-mining project, an ambitious effort to examine and compare structured and unstructured data, had been suspended after it was criticized in a Government Accountability Office report in March over the potential for misidentification. Another blow came when it was learned that pilot tests for the project had been using real data, which generated more criticism.

Homeland Security previously said the project would start up again in the fall, but acknowledged this week that it was ended, according to a story by the Associated Press. "The department has spent $42 million since 2003 developing the software tool," the AP said, and DHS' Science and Technology directorate "determined that new commercial products now offer similar functionality while costing significantly less to maintain than ADVISE."

My colleague Rob Preston wrote a column recently about the efficacy of data mining --or lack thereof -- based on a report by the Cato Institute, co-authored by one IBM's leading experts on information management, Jeff Jonas. Data mining isn't nearly accurate enough to be helpful in predicting terrorism, according to the report, especially in comparison to problems with misidentification.

Nonetheless, the federal government is committed to data mining as an anti-terrorism tool. ADVISE was just one of 12 data-mining projects going on at Homeland Security, according to AP. The FBI snuck $12 million into its most recent budget proposal for a new data mining project of its own.

That being the case, aren't there CIOs or database managers in the commercial sector who have hard-won expertise in data mining who could help out the federal government? Isn't that an appropriate response to the ongoing problems the feds are having using these advanced IT tools?