Offering New HOPE in the Balance Of Security and Civil Liberties
Data analytics gives law enforcement and intelligence agencies powerful tools that still protect privacy and civil liberties
December 2, 2008
We've all seen the movies and television shows where police detectives and anti-terrorist teams plug a suspect's name into a computer and receive all the relevant data they need to stop a crime or attack before it happens.
But the reality of law enforcement and intelligence work is not nearly as simple as these Hollywood fantasies appear. In a world that is drowning in data, where everyone from border guards to supermarket checkouts gather personal information, finding the right facts in thousands of disparate databases is no small feat.
For many people, even the possibility of such a search raises concerns that personal privacy and civil liberties could get lost even if the right information is found.
Into this delicate challenge comes DI-HOPE-KD, a suite of knowledge discovery tools developed at Rutgers University under the leadership of William M. Pottenger, a research professor of computer science.
Short for Distributed Interactive Higher-Order Privacy-Enhancing Knowledge Discovery, DI-HOPE-KD can take diverse sources of data, be they databases, news reports, or text documents and find intuitive associations, connections and links between them.
Law enforcement agencies have been doing this type of connect-the-dots work for decades, but it often requires weeks and sometimes months of painstaking formatting and reviewing by human eyes before useful data is discovered.
By looking for higher-order links--links that connect more than two dots--DI-HOPE-KD can do the same job in minutes and sometimes even seconds, saving a great deal of time and producing more useful and accurate information than traditional searches.
In an interview with the National Science Foundation, which is helping to fund the development of DI-HOPE-KD, Pottenger describes the work he and his team are doing to develop interactive privacy-enhancing characteristics, including those used in DI-HOPE-KD.
The framework is designed, Pottenger says, to keep a "human in the loop," that is, it depends on human intervention at several critical junctures, allowing privacy and other concerns to be more easily factored in as discovery progresses.
According to Pottenger, DI-HOPE-KD also allows different agencies and databases to collaborate and share information in an intuitive way without sharing all the specifics of that information. He gives the real-life example of a case where a framework like DI-HOPE-KD can tell investigators that a higher order link exists between a pseudoephedrine manufacturer in one town, a drug-dealing broker in another and an illicit meth lab in a third without revealing the details of the connection.
The investigators can then use other traditional means, including warrants and court orders, to pin down the nature of the connection and make arrests.
Pottenger believes there are far-reaching applications for this type of intuitive, yet privacy-enhancing data search, sharing and knowledge discovery technology in fields such as healthcare and retailing as well as in everyday tasks of information search and sharing.
He is also the CEO of Intuidex, a startup company working to commercialize DI-HOPE-KD, and is already collaborating with the Port Authority of New York and New Jersey and others to demonstrate the value of the technology in the field.-- Dana Cruikshank, National Science Foundation (703) 292-7738 email@example.com
William M. Pottenger describes the safeguards in creating technology to protect civil liberties.
Credit and Larger Version
William M. Pottenger, a research professor at Rutgers University and CEO of Intuidex.
Credit and Larger Version
Rutgers University New Brunswick
#0712139 III: Visual Analytics for Steering Large-Scale Distributed Data Mining Applications
#0534276 Collaborative Knowledge Discovery in Digital Government Data Using Distributed Higher-Order Text Mining
#0703698 Collaborative Knowledge Discovery in Digital Government Data Using Distributed Higher-Order Text Mining