The Intelligence Advanced Research Projects Activity has asked industry to propose methods to find speech and text content in low-resource languages relevant to domain-contextualized English queries.
IARPA said Thursday in a FedBizOpps notice the Machine Translation for English Retrieval of Information in Any Language program seeks an end-to-end system that can support monolingual triage of multilingual datasets.
The agency added it wants content identification methods that require minimal training data and have the capacity for rapid deployment to new languages and domains.
Participants in the MATERIAL program will work to create an “English-in, English-out” information retrieval system designed to recover data from a multilingual repository and present the information as summaries in English language.
IARPA will accept proposals from interested parties through March 20 and said the agency expects to kick off the program in October.