The Defense Advanced Research Projects Agency (DARPA) has announced plans to issue a new Broad Agency Announcement (BAA) titled Deep Exploration and Filtering of Text (DEFT) — a program that will require leveraging advances in machine learning, natural language processing, and computational linguistics “to assist warfighters with planning and decision-making by inferring implicit information in text, filtering redundancy and connecting like documents.” In anticipation of the BAA, DARPA plans to hold a Proposers’ Day this Wednesday, May 16th in Arlington, VA, to familiarize the community with its vision and goals for DEFT.

Here’s a description of the DEFT program (following the link):

The Deep Exploration and Filtering of Text (DEFT) program seeks to develop the ability to see through language to meaning in text, to make use of key information contained in text documents, to cue up information sources that contain new developments for analysts, and to automate the initial stages of report writing. The DEFT program aims to enable analysts to investigate orders of magnitude more documents and to discover implicitly-expressed, actionable information. This aim requires the development of automated deep natural language understanding technology. Technology developed in DEFT is expected to provide the capability to identify and interpret both explicit and implicit information from highly ambiguous and vague narrative text, and integrate individual facts into large domain models for assessment, planning and prediction. The goal of DEFT is to create an end-to-end system which will assist the warfighter and intelligence analyst at every stage — from searching through data to identifying threats and then formulating analyses. The DEFT System will enable analysts to move from limited linear processing of insurmountable amounts of data to a nuanced, strategic exploration of all available information.

According to Bonnie Dorr, the DARPA program manager for DEFT:

“Overwhelmed by deadlines and the sheer volume of available foreign intelligence, analysts may miss crucial links, especially when meaning is deliberately concealed or otherwise obfuscated. DEFT is attempting to create technology to make reliable inferences based on basic text. We want the ability to mitigate ambiguity in text by stripping away filters that can cloud meaning and by rejecting false information. To be successful, the technology needs to look beyond what is explicitly expressed in text to infer what is actually meant.

 

“Much of the basic research needed for DEFT has been accomplished, but now has to be scaled, applied and integrated through the development of new technology.”

To learn more, check out DARPA’s press release announcing the soon-to-be-released DEFT program — as well as details of this Wednesday’s Proposers’ Day.

And stay tuned as we’ll have more when the BAA is released.

(Contributed by Erwin Gianchandani, CCC Director)