Translation software helps scour documents for intelligence

A tremendous volume of documents ? both paper and electronic ? are captured during military and intelligence operations around the world. Many of those documents require translation into English. But sometimes the volume is so great there simply aren't enough linguists to analyze the documents.

"If you have a dearth of linguists, you may never look at some of these documents. There are a large volume of documents from Desert Storm that have not been translated. If you're looking for a needle in a haystack, the needle may still be there," said Doug Dearie, senior vice president and chief technology officer of Vredenburg. The Reston, Va., company creates document management systems for federal agencies.

Now, a new Vredenburg system is helping find the needles in the haystacks. With DOCEX ? for "document exploitation" ? the U.S. military and intelligence community can more quickly identify important documents so that vital information can be used during current operations, Vredenburg officials said. The solution is being rapidly deployed to support the military overseas, but company officials declined to say where it is being used.

"We are delivering this system more quickly than anticipated because of recent events," said Larry Den, Vredenburg's senior vice president of information technology.

"It's not a perfect science at this point," Den said, "but instead of waiting for the perfect thing, the community is saying 'Let's go with what we've got and make the best of that.'"

Two contracts funded development and deployment of DOCEX. One contract, worth $1.3 million over one year, was awarded through the Army CECOM R2 contract. The second contract, worth $793,000, was awarded by the Army Research Laboratory to support the rapid delivery of multiple HighView DOCEX suites to the field.

DOCEX is built on Vredenburg's HighView product suite for document management. It incorporates software tools developed by the Defense Department and private sector firms, including Vredenburg. It categorizes and translates large collections of documents while building a searchable repository. The system recognizes 187 languages.

While the translation isn't perfect, it is good enough at this point to help government officials determine if the documents have intelligence value. Then the linguists go to work, Dearie said.

"You are capturing this stuff on the ground and then you may need to exploit it tactically," he said.