Dina Mohamed - Purdue University Department of Linguistics
Students: Spring 2024, unless noted otherwise, sessions will be virtual on Zoom.
NL Downgrading, Sanitization, and Declassification
Dec 10, 2003
Abstract
Since Executive Order 12958, Classified National Security Information, signed by President Clinton on April 17, 1995, most U.S. Government agencies have faced a monumental problem of declassifying millions of pages of its documentation. Many agencies as well as other organizations and corporations are increasingly facing the need of downgrading or sanitizing information that they need or have to share with their various coalition partners, e.g., within the NATO alliance. Requirements of enhanced security as well as changing technologies and intelligence data make it necessary to revise existing criteria of declassification as well as to doublecheck that the libraries of presumably open documents do not indeed contain any sensitive information. It is equally important to assure that no such information is inadvertently revealed in e-mail and other online communications. Experience shows that doing all of this manually is prohibitively slow and expensive, so NL downgrading explores ways to apply NLP to the problem.About the Speaker
Dina Mohamed is a Ph.D. student in Linguistics, specializing in Natural Language Processing. She is the first recipient of the Purdue Graduate NLP Certificate. In 2000-2001, she was a member of the Atallah-Raskin CERIAS Research Group in NL IAS, where she co-worked with Dr. Raskin and (now Dr.) Christian \"Kiki\" Hempelmann on NL downgrading/sanitization/declassification, on which she also wrote her Masters\' theis in 2001. She has co-authored an article on syntactic NL watermarking with the other members of the group.
Ways to Watch
