U. to help track threats
But the program raises free-speech concerns
The University of Utah is participating in a major
research effort to help the Department of Homeland Security track
potential threats to the nation.
Critics, however, are raising concerns about free speech.
A new research program, to be developed along with Cornell University and the University of Pittsburgh, will teach computers to scan through text and sort opinion from fact. The three universities will comprise one of four University Affiliate Centers to conduct research that will contribute to national security.
The UACs will produce computational methods that have the potential to help the government detect negative opinions of the United States or its leaders. The program has been made possible by a $2.4 million grant from the DHS. The UACs will identify common patterns from numerous sources of information — such as newspapers and other publications — which may be indicative of potential threats to the country's security.
The researchers are eager to get started. But Dani
Eyer, executive director of the American Civil Liberties Union of Utah,
wonders why the DHS is sponsoring this kind of endeavor.
Eyer recalled an aborted effort in 2002 by the Pentagon to develop a tracking system called Total Information Awareness, intended to detect terrorists by analyzing information.
"In a post-9/11 climate and with ever-increasing technological capabilities, we sometimes get bigger monsters with even bigger chains," Eyer said.
There is also a federal law that prohibits DHS, or other intelligence agencies, from building such databases on American citizens. Researchers contend the programs they are developing will not be used for such things because they will be focusing on information, such as newspapers.
One of the worries the ACLU has, however, is the effect the project could have on someone who wants to write a letter to the editor. People may be more reluctant to do so because they fear becoming a government target.
"It is based on a faulty premise that terrorism can be prevented by collecting as much information as possible," Eyer said. "You don't find a needle in a haystack by bringing in more hay."
Researchers at the U., in the natural language processing (NLP) research group, will conduct research to improve information extraction technology. That will help computers recognize certain types of information, and extract facts and entities related to events.
"The goal of our research will be to develop new computational techniques to accurately identify facts associated with events," said Ellen Riloff, associate professor of computer science at the University of Utah.
For example, if the government wanted to create a system to detect possible terrorism, the information extraction (IE) technology would identify names of perpetrators, victims, physical targets, weapons, dates and locations.
"The U. has had an ongoing collaboration with NLP researchers at Cornell and the University of Pittsburgh," Riloff said. "We felt that our research expertise in natural language processing and our experience as collaborators made us particularly well-suited to submit a strong proposal."
As a result, the U. will receive $750,000 over a three-year period to conduct the research.
Another aspect that helped the university is that it already has programs under development for IE — programs like AutoSlog-TS, which generates a dictionary of extraction patterns, and Sundance, which focuses on information extraction of ungrammatical or ill-formed information. They will be used in the continued research.
Both undergraduate and graduate students will be directly involved with the research, and topics related to the research will be integrated into the educational programs at the universities, Riloff said.
In addition to the research project, the new UAC has educational goals, like training students to work in information extraction and presenting seminars and workshops for other researchers, according to Claire Cardie, a professor of computer science at Cornell.
The program also will emphasize recruitment of women and minorities.
"A substantial component of our University Affiliate Center will be aimed at increasing the representation of women and minorities in computer science," Riloff said. "We plan to reach out to women and minorities to increase their exposure to NLP technologies and their access to research opportunities."
The program to establish university-based research in information security and create information engineers is potentially renewable after three years.
"This effort will bring together an outstanding group of researchers with a proven track record in information analysis," Jeffrey W. Runge, acting under secretary for science and technology, said in a statement from the DHS. "The biggest challenge facing this critical area is the need for improved methods to quickly and accurately analyze, organize and make sense of vast amounts of changing data."
Critics, however, are raising concerns about free speech.
A new research program, to be developed along with Cornell University and the University of Pittsburgh, will teach computers to scan through text and sort opinion from fact. The three universities will comprise one of four University Affiliate Centers to conduct research that will contribute to national security.
The UACs will produce computational methods that have the potential to help the government detect negative opinions of the United States or its leaders. The program has been made possible by a $2.4 million grant from the DHS. The UACs will identify common patterns from numerous sources of information — such as newspapers and other publications — which may be indicative of potential threats to the country's security.
Eyer recalled an aborted effort in 2002 by the Pentagon to develop a tracking system called Total Information Awareness, intended to detect terrorists by analyzing information.
"In a post-9/11 climate and with ever-increasing technological capabilities, we sometimes get bigger monsters with even bigger chains," Eyer said.
There is also a federal law that prohibits DHS, or other intelligence agencies, from building such databases on American citizens. Researchers contend the programs they are developing will not be used for such things because they will be focusing on information, such as newspapers.
One of the worries the ACLU has, however, is the effect the project could have on someone who wants to write a letter to the editor. People may be more reluctant to do so because they fear becoming a government target.
"It is based on a faulty premise that terrorism can be prevented by collecting as much information as possible," Eyer said. "You don't find a needle in a haystack by bringing in more hay."
Researchers at the U., in the natural language processing (NLP) research group, will conduct research to improve information extraction technology. That will help computers recognize certain types of information, and extract facts and entities related to events.
"The goal of our research will be to develop new computational techniques to accurately identify facts associated with events," said Ellen Riloff, associate professor of computer science at the University of Utah.
For example, if the government wanted to create a system to detect possible terrorism, the information extraction (IE) technology would identify names of perpetrators, victims, physical targets, weapons, dates and locations.
"The U. has had an ongoing collaboration with NLP researchers at Cornell and the University of Pittsburgh," Riloff said. "We felt that our research expertise in natural language processing and our experience as collaborators made us particularly well-suited to submit a strong proposal."
As a result, the U. will receive $750,000 over a three-year period to conduct the research.
Another aspect that helped the university is that it already has programs under development for IE — programs like AutoSlog-TS, which generates a dictionary of extraction patterns, and Sundance, which focuses on information extraction of ungrammatical or ill-formed information. They will be used in the continued research.
Both undergraduate and graduate students will be directly involved with the research, and topics related to the research will be integrated into the educational programs at the universities, Riloff said.
In addition to the research project, the new UAC has educational goals, like training students to work in information extraction and presenting seminars and workshops for other researchers, according to Claire Cardie, a professor of computer science at Cornell.
The program also will emphasize recruitment of women and minorities.
"A substantial component of our University Affiliate Center will be aimed at increasing the representation of women and minorities in computer science," Riloff said. "We plan to reach out to women and minorities to increase their exposure to NLP technologies and their access to research opportunities."
The program to establish university-based research in information security and create information engineers is potentially renewable after three years.
"This effort will bring together an outstanding group of researchers with a proven track record in information analysis," Jeffrey W. Runge, acting under secretary for science and technology, said in a statement from the DHS. "The biggest challenge facing this critical area is the need for improved methods to quickly and accurately analyze, organize and make sense of vast amounts of changing data."
E-MAIL: blee@desnews.com
