Machine Learning and Credit-Taking: An Inductive Approach to Disentangling Terrorist Attention-Seeking Behavior

Publikation: Working paperForskning

Abstrakt

Most academic definitions of terrorism emphasize the communicative function of terrorism. The aim of terrorist violence is to gain public attention to a political or religious cause. Despite this emphasis on communication, terrorists rarely seek attention by claiming responsibility for attacks. According to the Global Terrorism Database claims of responsibility are only issued for approximately every fifth attack. This begs the question: why do terrorists abstain from the easy ‘win’ of claiming their attacks? Previous research has theorized that factors like state sponsorship, principal-agent problems, casualty-levels, and inter-group competition are important factors in explaining variation in terrorist credit-taking propensities. In this paper, we diverge from the tried and trusted deductive approach and instead utilize an inductive approach. We apply machine learning techniques using tree-based ensemble methods to predict absence of claims on out-of-bag observations. Our initial results indicate that geographical factors are the strongest predictors of the absence of claims - something largely overlooked by existing research.
OriginalsprogEngelsk
StatusUnder udarbejdelse - 2020

Fingeraftryk Dyk ned i forskningsemnerne om 'Machine Learning and Credit-Taking: An Inductive Approach to Disentangling Terrorist Attention-Seeking Behavior'. Sammen danner de et unikt fingeraftryk.

  • Citationsformater