REAL TIME BIG DATA ANALYTICS FOR PREDICTING TERRORIST INCIDENTS

Author/Creator

Author/Creator ORCID

Date

2017-01-01

Department

Information Systems

Program

Information Systems

Citation of Original Publication

Rights

This item may be protected under Title 17 of the U.S. Copyright Law. It is made available by UMBC for non-commercial research and education. For permission to publish or reproduce, please see http://aok.lib.umbc.edu/specoll/repro.php or contact Special Collections at speccoll(at)umbc.edu
Distribution Rights granted to UMBC by the author.

Abstract

Terrorism is a complex and evolving phenomenon. In the past few decades, we have witnessed an increase in the number of terrorist incidents in the world. The security and stability of many countries is threatened by terrorist groups. Perpetrators now use sophisticated weapons and the attacks are more and more lethal. Currently, terrorist incidents are highly unpredictable which allows terrorist groups to attack by surprise. The unpredictability of attacks is partly due to the lack of real time terrorism data collection systems, adequate risk models, and prediction methodologies. To bridge the gap between terrorist incidents and counter-terrorism measures, it is crucial to develop real time terrorism data collection systems along with novel and proven risk models and prediction methodologies. In this research, we developed a set of systems and methodologies to collect and analyze terrorism related data. The methodologies include terrorism data summarization for root cause analysis, cluster analysis of terrorist attacks into groups with similar patterns, a novel risk model that uses data collected via our data collection system, and a prediction method that uses our risk model and Markov Chains. Our methodology for root cause analysis utilizes our novel algorithm, along with Latent Dirichlet Allocation and historical terrorism data of START. Our clustering method segregates terrorist groups based on their similarities in attack patterns. Our data collection system is an automated crawler engine that collects data from selected data sources via RSS Feeds and on demand. The result is real time data that gets preprocessed automatically using selected keywords. Our novel terrorism risk model utilizes the preprocessed data to calculate terrorism risk levels at different locations. Lastly our prediction method utilizes our terrorism risk model, and Markov Chain models to predict future terrorist incidents in different countries. We have implemented a fully automated system that does not require any manual interventions for collecting and calculating the risk values. The results obtained in this research show a promising terrorism prediction system which predicts future attacks up to three months prior to the occurrence of an attack with a maximum of 96.85% precision and 96.32% recall. Our software system and methodologies can be a useful tool for terrorism analysts to improve counter-terrorism measures, and potentially prevent future terrorist attacks.