Events, Relationships, and Script Learning for Situational Awareness
October 2017 • Poster
Poster for research into using machine learning to extract patterns from high volumes of textual data
Intelligence analysts working on new tasks have difficulty extracting meaningful information and recognizing patterns from high volumes of open and closed source textual data. Their efforts suffer from the absence of automated methods that reliably recognize actors, activities, and objects comprising an event or sequence of events (i.e., scripts). Further, the state-of-the-practice lacks a means to transfer knowledge learned from previously identified situations to new situations. In the project, we address those deficiencies to provide first steps towards a system that recognizes events, sequences of events, instances of scripts and that supports transfer of this information to new situations.