IARPA needs help identifying ‘emerging events’

Text Mining Intelligence Community analysts have tried in the past to identify emerging and evolving events from the documents they are examining, but that goal has proved to be very elusive.

“Some technologies have tried to identify unnamed events in data (i.e., those that represent actions or activities that are unfolding) by identifying all the verbs in a document but relating those ‘events’ to one another and merging ‘events’ across documents have met with little success,” explains the Intelligence Advanced Research Projects Activity (IARPA), in a request for information it issued on Feb. 18.

IARPA’s RFI seeks input from companies and organizations that might have new ideas on how to approach this challenge of spotting “emerging events” and identifying “participating entities.”

The input it receives may help IARPA shape a broad agency announcement it anticipates releasing on this topic, and the input could also influence the agenda of a one-day workshop IARPA expects to host in the Washington, DC area in June. “An expected result of such a workshop is the identification of promising areas for investment through vehicles such as seedlings, grand challenges, and programs,” says IARPA.

IARPA wants to figure out better ways to track event emergence as reported in unstructured text.

“Techniques must move beyond verb-based event definition and extraction to other, richer characterizations of events and to approaches that find and link events as higher-level events emerge,” IARPA explained.

Interested parties are invited to submit their capabilities and ideas by March 28. IARPA welcomes a one-page cover sheet, half-page executive summary and description of the suggested approaches in no more than five pages.

Further information is available from Heather McCallum-Bayliss, a program manager, at dni-iarpa-rfi-14-05@iarpa.gov.