Design of an Ensemble Machine Learning Based Recommender System Framework For Terrorism Prevention

Abstract

Terrorists cause a lot of unrest, fear and destruction of live and property worth trillions of Naira in Nigeria and the entire world. Several authors had deployed hard computing techniques like kinetic approach to prevent and combat the heinous act but the menace kept increasing. Hence, there is need for deployment of soft computing techniques such as machine learning to combat the problem of terrorism. This study created a machine learning method to forecast terrorist activity and warn the public and security organizations so they can take preventative action. The paper proposes bagging techniques, consisting of the traditional ensemble module (logistic regressing, random forest and support vector machine) and deep learning module (bidirectional long short-term memory and bidirectional encoder representation from transformer) to explore both the global terrorist dataset (GTD) and dataset obtained from social media platform for predicting the likelihood of the terrorist attack, the likely time and possible location of future attack