DARPA in search of universal translator tech
In their military, diplomatic and heath care roles, U.S. government representatives frequently encounter low-resource languages, which often are less-studied, less-privileged or spoken by small numbers of people such as Inuit or Sindhi.
For many such languages, no available automated language technology exists. Translation resources are expensive and tend to be focused on the most popular languages. But for more than 7,000 languages in the world, military, diplomats and health workers need a way to quickly and inexpensively communicate.
The Defense Advanced Research Projects Agency hopes to address that challenge with its Low Resource Languages for Emergent Incidents (LORELEI) program, which aims to advance computational linguistics and language technology for low-resource languages.
According to a notice in FedBizOpps, DARPA wants to develop language technology that leverages universal language principles instead of relying on huge, manually-translated, transcribed or annotated texts. It also aims to deliver situational awareness for the low-resource foreign language sources as soon as 24 hours after a new language requirement emerges.
While LORELEI technologies may include partial or fully automated speech recognition and/or machine translation, the overall goal is not to be translating foreign languages into English. Instead, the systems would provide situational awareness by identifying elements of information in foreign language and English sources, such as topics, names, events, sentiment and relationships.
DARPA is holding a proposer’s day on Nov. 13 to provide information to potential responders to the anticipated LORELEI broad agency announcement.
Posted by GCN Staff on Oct 29, 2014 at 10:11 AM