Here’s how intelligence agencies can search foreign documents without learning the language

WASHINGTON – The intelligence group now has a software that enables English-speaking customers to search by way of foreign language textual content and speech for data.The new software was developed by Raytheon BBN Technologies in partnership with the Intelligence Advanced Research Projects Activity — a company inside the Office of the Director of National Intelligence that develops applied sciences to unravel a few of the intelligence group’s hardest issues.Essentially, as soon as English-speaking customers enter a search question in English, the program appears to be like by way of foreign language documents and recordings to seek out related outcomes, translating these phrases again into English earlier than presenting outcomes again to the person. It’s an “English-in, English-out” software, and the firm claims its system permits operators to search foreign documents, discover outcomes and perceive their context and that means without having to talk the language, based on a Jan. 31 announcement.Raytheon mentioned they used Kazakh, Pashto, Somali, Swahili and Tagalog as the low knowledge foreign languages for its machine learning algorithm, which was moreover examined towards Farsi, Bulgarian, Lithuanian and Georgian.“The system is designed to be utilized to any foreign language,” mentioned Raytheon BBN Program Manager John Makhoul in an announcement. “Low-resource languages current a specific problem to retrieval and translation applied sciences due to an absence of knowledge for coaching methods. Raytheon BBN met that problem by growing strategies to beat the concern of low knowledge and utilized them to an end-to-end system that exceeded the objectives of the program.”The answer is a part of IARPA’s Machine Translation for English Retrieval of Information in Any Language, or MATERIAL, program, launched in 2017. Raytheon is one in all 4 prime contractors growing options, together with Johns Hopkins University, Columbia University and the University of Southern California Information Sciences Institute. Each vendor was given a bundle of coaching knowledge to develop machine learning options. MIT Lincoln Laboratory, the University of Maryland Center for Advanced Study of Language, the National Institute of Standards and Technology and Tarragon Consulting made up the take a look at and analysis crew that assessed performances.“The instruments and strategies developed beneath the program will enhance our skill to seek out, look at and analyze foreign language content material without needing to be taught the language,” mentioned IARPA MATERIAL Program Manager Carl Rubino in an announcement. “For low-resource languages the place experience is minimal, these new capabilities present a major benefit.”Raytheon Technologies didn’t disclose the worth of its contract with IARPA, though it famous that it has been working with the group on the answer for 4 years. In an announcement at the time of the preliminary awards, Columbia University mentioned its grant was $14 million.Nathan Strout is the workers editor at C4ISRNET the place he covers the intelligence group.

https://www.c4isrnet.com/intel-geoint/2022/01/31/heres-how-intelligence-agencies-can-search-foreign-documents-without-learning-the-language/

Recommended For You