Communications of the ACM
An Extra Challenge in a Crisis: Dealing With Rare Languages
Research director Kevin Knight (seated) with USC Information Sciences Institute translation team members (standing from left) Thamme Gowda, Sasha Mayn, and Jonathan May.
Credit: Caitlin Dawson / USC
Researchers at the University of Southern California's Information Sciences Institute (ISI) are building machine-learning systems to quickly translate any of the world's 7,000 languages and automatically generate actionable information.
The team participated in an assignment to translate two Ethiopian languages currently unknown to machine-translation systems as part of a U.S. Defense Advanced Research Projects Agency initiative to develop a rapid automated language toolkit for languages currently missing from linguistic databases that feed online translation systems. Their approach involved a program that converts any language's writing system into the Latin-based alphabet; a name-finding tool highlighting the names of people, places, and organizations to contextualize the document and identify what supplies are needed most and where; and a known language as an intermediate step for decoding similar words.
The ISI team says their work could be a step toward a universal translator for use in any region where international disaster relief teams have little language fluency.
From USC News
View Full Article
Abstracts Copyright © 2017 Information Inc., Bethesda, Maryland, USA
No entries found