Using Positive and Negative Network Ties in AI Models: An Application Predicting Terrorism in India

Abstract An effective coalition operation needs to consider both warfare with national adversaries as well as unorganized militias that may counter more technologically advanced adversaries through terrorism. To ensure the attainment of objectives, the commander needs to understand factors that prompt violent events. One way to understand these threats is to explore network ties between different regions and groups, where positive and negative ties may promote or reduce terrorism. Positive ties are edges that promote nodes to become similar to each other, or homophilous, while negative ties are edges that promote nodes to become dissimilar. We study the impact these ties may have on the ability to predict the occurrence of terrorist incidents in a district based on data about terrorist incidents in India from 2007-2018. Negative and Positive ties among districts are determined by comparing their demographic data, and this information is used to create an AI model that can predict the probability of violent occurrences in specific districts. We evaluate the results obtained by creating and comparing the predictive power of AI models that make predictions with and without positive/negative ties.
  • Dave Braines (IBM UK)
  • Diane Felmlee (PSU)
  • Dinesh Verma (IBM US)
  • Scott Sigmund Gartner (PSU)
Date Sep-2020
Venue 4th Annual Fall Meeting of the DAIS ITA, 2020