Reading machine to snoop on Web

(Credit: DARPA)

What if the wisdom of Web could be yours, without having to read through it one page at a time? That's what the military wants.

DARPA has hired a company to develop a reading machine to reduce the gap between the ever increasing mountain of digitized text and the intelligence community's insatiable appetite for data input.

BBN Technologies was awarded the $29.7 million contract to develop a universal text engine capable of capturing knowledge from written matter and rendering it into a format that artificial intelligence systems (AI) and human analysts can work with. (PDF)

The military will use the Machine Reading Program, as it's officially called, to automatically monitor the technological and political activities of nation states and transnational organizations-which could mean everything from al-Qaeda to the U.N.

To pull it off, BBN will "develop techniques that can generalize across the linguistic structure and content of diverse documents to extract relations and axioms directly from text rather than relying on a knowledge engineer to encode such information."

"The machine reading system that DARPA envisions is not evolutionary, but revolutionary," said BBN Technologies VP Prem Natarajan. "Such a system could eliminate many of the impediments to stability that our military faces such as a lack of understanding of local customs, and give us the ability to assess global technology developments continuously."

However, BBN also expects the program to enable a plethora of new civilian applications, everything from intelligent bots to personal tutors. The system could provide unprecedented access and automated analysis of the world's libraries, allowing for vastly expanded cultural awareness and historical research, according to the Cambridge, Mass.-based company.

BBN already offers a broadcast monitoring system that automatically transcribes real-time audio stream and translates it into English, creating a continuously updated, searchable archive of international television broadcasts.

"Imagine if the Reading System could be applied to scouring the World Wide Web for good deals on cars one time, and then applied to integrating new findings in genetics to an automated theory of disease," DARPA posits in its bid solicitation.

It should also be able to crank out one a heck of a term paper.