August 7, 2008

Scientists Explore Ways To Improve Data Analysis

Georgia Tech University announced Wednesday it had received a $3 million grant by the Department of Homeland Security and the National Science Foundation for the establishment of a new research field in visual and data analytics. 

And although researchers are highly optimistic about the nascent science, they must first find a more effective way to mine the enormous amounts of data from the Internet and complex scientific instruments.

As it turns out, even sophisticated computers can suffer from information overload. Such systems are capable of performing millions of calculations per second, however the task of analyzing the resulting mountains of data presents a challenge.

Scientists are now exploring new ways to sift through this vast amount of information, and transform it into meaningful data that researchers, health officials and even law enforcement can act on.

"We're looking at fundamental science, fundamental mathematics that in many ways are a mess of jumbled data," John Stasko, a Georgia Tech professor of interactive computing, told the Associated Press.

"We try to give them a structure, because as humans we make these inferences so much better when our data has structure."

For example, some researchers may analyze reports of medical conditions and drug purchases in response to disease outbreaks.

Stasko said some of his work will center on how the new data analysis methods might be used to crack down on criminals.  Witness statements, satellite images and other clues could be entered into a novel program that authorities might use to identify new leads, he said.

But however innovative the new programs, they will ultimately still need humans to transform the data into actionable results.  

"An investigator needs to put this all together and connect the dots, find the coherent story," Stasko said.


Image Caption: Carsten Goerg (pictured), a postdoctoral fellow in Georgia Tech professor John Stasko's research group in the School of Interactive Computing, analyzes a document collection using a visual analytics system called Jigsaw. (Georgia Tech Photo: John Stasko)


On the Net: