Previous Work: Extracting Event Attributes from Unstructured Textual Data for Persistent Situational Awareness

Principal Investigator(s): 
Collin Baker

In this collaborative project with Decisive Analytics Corporation (DAC), FrameNet researchers are developing semantic frames for representing the attributes of complex events, which permit more fine-grained analysis than other event recognition frameworks. The researchers are developing event recognition methods focused on organizations and how they plan and carry out actions. These methods are broadly applicable to actions planned and carried out by all types of organizations, such as corporations, government agencies, military units, and insurgent groups. 

The overall goal of the project is to develop state-of-the-art methods for automatically recognizing events and their attributes in unstructured text. This research will enhance FrameNet’s coverage of event attributes by developing new frames and adding to existing frames as necessary to analyze language describing organizations and the actions they plan and execute. These extensions will cover:

  • The semantics of all the standard modal verbs, e.g.:
    • Bill should have driven here. => He didn't drive here.
    • Bill should be here by now => Since he’s not, something must be wrong.
    • Bill would have been here by now. => If something had been different.
  • Negation (Bill didn't drive across the country.  None of them drove home that night.)
  • Distinguishing generic predication from specific (Bill drove a Chevy for years vs. Bill drove a Chevy to the party)
  • Conditionals:
    • If it rains, we’ll meet in the gym.
    • We’ll meet on the field unless it rains. 
    • In case it rains, the gym is open.

All of these features are needed to describe the alternative branches resulting from choices of action in a complex scenario. This capability will dramatically improve information management and analysis capabilities.