Homeland data mining efforts will differ from Pentagon's

The Homeland Security Advanced Research Projects Agency (HSARPA) is researching ways to mine large amounts of data, but its work will differ from that of a Defense Department agency that had one project killed because of privacy concerns, according to HSARPA's director.

The Defense Advanced Research Projects Agency (DARPA) made news last year for its Total Information Awareness (TIA) project, which called for technologies to search commercial databases in order to identify potential terrorists. HSARPA Director David Bolka said his agency will research data mining but, unlike DARPA, will not seek to mine individuals' data.

"The concept of data mining is one of those things that applies across the spectrum, from business looking at financial data to scientists looking at scientific data," Bolka said in a recent interview. "The application of data mining -- which databases do you look at, and what data do you extract from them, and for what purpose -- was where TIA got into trouble and the perceived violation of privacy, particularly the Privacy Act."

"We're not doing any of that stuff," Bolka added.

But he said the Homeland Security Department will mine data for information from biological sensors, for example. The purpose would be to determine "background" counts for the amounts of pathogens such as anthrax in populated areas, he said.

"Until we get a dense enough sensor network out there, we won't be able to find out," Bolka said. "Once we do get a dense enough sensor network out there, we are going to be inundated with data and a lot of the data mining techniques that have been used in industry ... particularly the financial one, will be applied to those data sets."

The practice will differ from DARPA's, Bolka said, because while the research may involve people, it likely will not be on the individual level. Instead, the sensor data on a pathogen's exposure in a certain area could be matched with demographic data to determine, for instance, the cross-section of people that probably have been exposed.

"Can you match it to individual data? Probably not," Bolka said. "And that's where TIA got into trouble because they were looking at compilations of data on individuals."

Bolka acknowledged that Homeland Security's directorate on information analysis and infrastructure protection, which has a mandate to analyze data for anti-terrorism purposes, could request that his office conduct data-mining research. But he said any technology that the private sector develops through his office would be subject to national laws and regulations.

"Technology is not inherently good or bad; it is a technology," Bolka said. "The application is subject to the laws, rules and regulations of the republic. That's not my job."

If lawyers, lawmakers and others determine that a particular technology application violates the law, he said, "then obviously we won't do it."