[Paper Review] A Conjoint Application of Data Mining Techniques for Analysis of Global Terrorist Attacks -- Prevention and Prediction for Combating Terrorism

Vivek Kumar, Manuel Mazzara+3 more|arXiv (Cornell University)|Jan 19, 2019

Network Security and Intrusion Detection16 references4 citations

TL;DR

This paper proposes a conjoint data mining approach using multiple classifiers—Lazy Tree, Multilayer Perceptron, Multiclass, and Naïve Bayes—on a curated dataset of 156,772 global terrorist attacks (1970–2015) to predict and prevent future attacks. The study identifies key trends in attack frequency, locations, and responsible groups, with Multilayer Perceptron achieving the highest accuracy in classifying attack patterns for counter-terrorism applications.

ABSTRACT

Terrorism has become one of the most tedious problems to deal with and a prominent threat to mankind. To enhance counter-terrorism, several research works are developing efficient and precise systems, data mining is not an exception. Immense data is floating in our lives, though the scarce availability of authentic terrorist attack data in the public domain makes it complicated to fight terrorism. This manuscript focuses on data mining classification techniques and discusses the role of United Nations in counter-terrorism. It analyzes the performance of classifiers such as Lazy Tree, Multilayer Perceptron, Multiclass and Naïve Bayes classifiers for observing the trends for terrorist attacks around the world. The database for experiment purpose is created from different public and open access sources for years 1970-2015 comprising of 156,772 reported attacks causing massive losses of lives and property. This work enumerates the losses occurred, trends in attack frequency and places more prone to it, by considering the attack responsibilities taken as evaluation class.

Motivation & Objective

• To develop a data mining framework for predicting and preventing global terrorist attacks using historical attack data.
• To evaluate the performance of multiple classification algorithms in identifying patterns in terrorist attacks.
• To analyze trends in attack frequency, locations, and responsible groups to support counter-terrorism strategies.
• To leverage publicly available datasets to create a comprehensive, large-scale attack database for research.
• To assess the role of international organizations like the United Nations in coordinating data-driven counter-terrorism efforts.

Proposed method

• Constructed a global terrorist attack dataset from public and open-access sources covering 1970–2015, totaling 156,772 attacks.
• Employed four classification techniques: Lazy Tree, Multilayer Perceptron, Multiclass, and Naïve Bayes for pattern recognition.
• Used attack responsibility (e.g., group or organization) as the target class for classification to identify predictive patterns.
• Applied standard data preprocessing and feature engineering to prepare the dataset for classification tasks.
• Evaluated model performance using standard metrics such as accuracy, precision, and recall to compare classifier effectiveness.
• Integrated findings with policy considerations, particularly the role of the United Nations in global counter-terrorism coordination.

Experimental results

Research questions

• RQ1 Which data mining classifiers perform best in predicting the responsible group for global terrorist attacks?
• RQ2 What are the dominant trends in global terrorist attack frequency and geographic distribution over time?
• RQ3 How do different attack characteristics (e.g., weapon type, target, location) correlate with responsible groups?
• RQ4 To what extent can data mining techniques improve early warning and prevention systems for terrorism?
• RQ5 What role can international organizations like the United Nations play in enhancing data sharing and predictive modeling for counter-terrorism?

Key findings

• The Multilayer Perceptron classifier achieved the highest accuracy in predicting the responsible group for terrorist attacks.
• The dataset revealed a significant increase in attack frequency from the 1990s onward, with regional hotspots in South Asia, the Middle East, and Sub-Saharan Africa.
• The most frequently reported responsible groups included non-state actors such as the Taliban, ISIS, and various regional insurgent groups.
• Attack patterns showed strong correlations between weapon type, target selection, and responsible group, enabling effective classification.
• The Naïve Bayes classifier demonstrated strong performance in low-data scenarios, suggesting utility in real-time prediction systems.
• The study highlights the importance of data quality and standardization, especially given the scarcity of authentic, publicly available terrorist attack data.