DEFT – Deep Learning Natural Language Processor from DARPA


Deep learning has made solid progress into 2014. A lot of research continues in natural language processing via this approach and DARPA is making significant contributions towards viable solutions. Its $25 million XDATA project is doing this through various frameworks, graph analysis and advanced visualization techniques among them. And as the big data collected by defense agencies continues to grow, there is a need for it to be organized and analyzed in order to extract useful information.

The Deep Exploration and Filtering of Text project launched in 2012 aims to achieve this goals for bodies of text as well as speech. DEFT is being developed to understand the semantics of text and therefore analyze ambiguous statements at a faster rate than humans. This will be packaged into alerts which can be flagged for suspicious messages. Other features expected in the project include creating a data store to help build correlations between things and people in the text.


Comments are closed.