Deep Learning Models for Predicting Globally Disruptive Events

Abstract

Disruptive events across the globe present significant challenges to United States foreign policy and military posture. Disruptive events, crises, or rising hotspots cover a range of events, such as unexpected changes in government, armed conflicts, terrorist attacks, and conflicts between governments and local populations. New global threats have increased instability, posing new challenges to policy development, escalation management, resource allocation and crisis response. Response strategies informed by evidence based research are increasingly needed to protect US interests. We propose to develop a method to forecast globally disruptive events by integrating social discourse information into a deep learning framework. We will both derive social discourse features from social media, and develop a deep learning model to integrate historical and social context with social discourse information. We will validate our approach by answering three research questions.The result of this project will yield a transformative ability to anticipate globally disruptive events. Early warnings increase the array of power projection, influence and escalation management options available to policy makers. Insights from this work can support existing theories, help shape new ones, and allow us to generalize our understanding of crisis antecedents across regions. Success will make it easier for the US and its allies to identify the best strategy for a situation, and provide sufficient opportunity to execute the strategy to manage a developing crisis.Planning for possible disruptive events relies on insights from a range of sources, but struggles to capture information about population behaviors. Our project provides a transformative capability that utilizes social discourse features derived from social media data to predict when globally disruptive events will occur. Success would enhance the process of planning US policy responses, and give US agencies and US allies critical lead time to prepare for possible disruptive events. Additionally, beyond a raw prediction, our system can be used to provide much needed context to an analyst, calling attention to emerging situations much sooner than would otherwise be noticed.

Document Details

Document Type
DoD Grant Award
Publication Date
May 23, 2019
Source ID
N000141912316

Entities

People

  • Anna Buczak

Organizations

  • Johns Hopkins University
  • Office of Naval Research
  • United States Navy

Tags

Readers

  • Strategic Security Studies
  • Systems Analysis and Design

Technology Areas

  • AI & ML
  • AI & ML - DoD AI Strategy
  • AI & ML - Neural Networks