Big data-based prediction of terrorist attacks

doi:10.1016/j.compeleceng.2019.05.013

Computers & Electrical Engineering

Volume 77, July 2019, Pages 120-127

https://doi.org/10.1016/j.compeleceng.2019.05.013 Get rights and content

Abstract

An optimised hybrid classifier is proposed for the prediction of terrorist attacks. Hybrid classifier is designed using big data. It puts forward a framework that includes data collection, preprocessing, hybrid classification mining, and classifier testing as a single unit in predicting the terrorist attacks. The genetic algorithm is used to optimise the weight of each single classifier to improve the prediction accuracy of the hybrid classifier. The results reveal that the hybrid classifier is superior to the single classifier in prediction accuracy.

Introduction

The shadow of terrorist attacks has loomed globally since the 9–11 attacks. Although countries have stepped up their efforts to prevent and control them, terrorist attacks have not been far from people's lives. The Global Terrorism Database (GTD) shows that a total of 25,903 terrorist attacks took place around the world between 2000 and 2012; the average is approximately 2000 per year and more than 5 times per day [1]. There are many underlying correlations and laws in the numbers associated with terrorist attacks. If these hidden phenomena can effectively guide the construction of anti-terrorism early warning systems, they can help solve the difficult problems in counter-terrorism decision-making and conduct regular identification and prediction. This has become a research topic in many fields of informatics, as it improves the risk management and accurate warning of terrorist attacks through the use of classification technology to mine large data about terrorist attacks in depth and minimise the unknown risk of terrorist attacks.

The University of Maryland built the GTD as an open-source collection that includes data sets from the 1970s to 2014 on global terrorist activity. The database has collected and combed approximately 140,000 terrorist incidents worldwide from 1970 to 2014 and includes information about the timing, location, use of weapons and targets, number of casualties, and identifiable responsible parties. The average terrorist attack contains up to 45 messages, where the one with the most has more than 120. GTD provides researchers with comprehensive, reliable, and open-source data, and helps uncover the underlying structure behind terrorist attacks. For example, reference [2] is a data source based on the GTD's listed terrorist attacks in India that analyses the type and number of attacks related to the attack organisation. Reference [3] divides the terrorist attacks in GTD into transnational events and domestic events, analyses and compares the impact of the both, and finds that transnational terrorism will bring more negative impact on a country's economic growth than domestic terrorism. The research results also show that domestic terrorism can evolve into transnational terrorism. The target country cannot ignore the domestic terrorism of its neighbours in order to reduce the risk of terrorist attacks in the country, and it should cooperate to eliminate terrorism with its neighbours. Literature [4] uses GTD as a data source, a visual environment is introduced to explore various types of terrorist attacks, including attack target type and attack strategy.

The paper is organised as follows: Section 2 discusses the research methods, i.e., data collection, preprocessing, and construction of the classifier. The analysis of the experiments conducted is presented in Section 3. Section 4 presents the conclusion.

Section snippets

Research methods

In this paper, a large-data prediction framework for terrorist attacks based on a mixed classifier is presented (see Fig. 1). The framework includes four steps: data collection, data preprocessing, construction of the mixed classifier, and data testing [5], [6], [7], [8]. In the data-collection phase, the framework learns about the types of attacks and attributes that GTD contains. In addition, more than 140,000 terrorist attacks in the database were examined in terms of the number of terrorist

Analysis and discussion of experimental results

The iterative curve of GA optimisation integrated classifier is presented in Fig. 8 to verify the convergence of the proposed algorithm in this paper. It can be observed that the optimal value of the population shows a step-down trend. Furthermore, its convergence speed is less than 30 times, and its prediction error rate is stable at 6%. The population mean curve fluctuates between 7.5% and 8.5%. The prediction accuracy is still higher than that of the optimal single classifier (bagging 91.5%)

Conclusion

Terrorists attacks have been increasing in recent times, and prediction of these attacks is almost impossible. The GTD was used as big data, and hybrid classifiers were trained to predict new attacks. The data, after preprocessing for noise, was categorised based on the type of attacks. Then, the hybrid classifier with KNN, decision tree C4.5, bagging, and SVM was designed with a GA to predict attacks. Experiments were also conducted for individual classifiers without the hybrid approach. From

Conflict of interest

None.

Meng Xi is currently a lecturer in the department of Crime Investigation and Counterterrorism of the People's Public Security University of China. Her research interest is mainly in the area of information visualization, counterterrorism, and semantic analysis. She has published several research papers in scholarly journals in the above research areas and has hosted two ministry-level projects.

References (25)

A.A. Mutlag et al.
Enabling technologies for fog computing in healthcare IoT systems
Futur Gener Comput Syst
(2019)
G LaFree et al.
Introducing the global terrorism database
Terror Political Violence
(2007)
A Khan et al.
Terrorism and India: an economic perspective
Qual Quant
(2016)
W Enders et al.
Domestic versus transnational terrorism: data, decomposition, and dynamics
J Peace Res
(2011)
D Guo et al.
Visualizing patterns in a global terrorism incident database
Environ Plann B Plan Des
(2007)
KM Leung
Naive Bayesian classifier
(2007)
J J Hopfield
Artificial neural networks
IEEE Circuits Devices Mag
(1988)
R. Kohavi
Scaling up the accuracy of Naive-Bayes classifiers: a decision-tree hybrid
J Chen et al.
An improved ID3 decision tree algorithm
S Nizamani et al.
Detecting terrorism incidence type from news summary
(2012)

S Shafiq et al.

Attack type prediction using hybrid classifier

N Stepenosky et al.

Majority vote and decision template based ensemble classifiers trained on event related potentials for early diagnosis of Alzheimer's Disease

Int J Comput Appl

(2006)

Cited by (22)

External blast flow field evolution and response mechanism of single-layer reticulated dome structure
2023, Defence Technology
Citation Excerpt :
Statistics of global terrorism database (GTD) reveals that nearly 200,000 incidents of terrorist bombing have taken place in 1970–2019 [1], and in recent times there are on average 2000 incidents per year (i.e., more than five per day) [2].
Single-layer reticulated dome structure are commonly high-profile building in the public and can be attractive targets for terrorist bombings, so the public can benefit from enhanced safety with a stronger understanding of the behavior of single-layer reticulated dome structure under explosion. This paper investigates the fluid-structure interaction process and the dynamic response performance of the single-layer reticulated dome under external blast load. Both experimental and numerical results shown that structural deformation is remarkably delayed compared with the velocity of blast wave, which advises the dynamic response of large-span reticulated dome structure has a negligible effect on the blast wave propagation under explosion. Four failure modes are identified by comparing the plastic development of each ring and the residual spatial geometric of the structure, i.e., minor vibration, local depression, severe damage, and overall collapse. The plastic deformation energy and the displacement potential energy of the structure are the main consumers of the blast energy. In addition, the stress performance of the vertex member and the deep plastic ratio of the whole structure can serve as qualitative indicators to distinguish different failure modes.
A two-step machine learning method for casualty prediction under emergencies
2022, Journal of Safety Science and Resilience
Citation Excerpt :
The subsequent explosion killed 15 people and injured at least an additional 151. To reduce the risk of emergencies to people and society, risk prediction and assessment have been widely used [1–8]. For instance, Paté-Cornell and Guikema [7] proposed a risk analysis model based on Bayesian probability, which further contributed to establishing a decision-making model for setting priorities among threats and countermeasures.
Casualty prediction is meaningful to the emergency management of natural hazards and human-induced disasters. In this study, a two-step machine learning method, including classification step and regression step, is proposed to predict the number of casualties under emergencies. In the classification step, whether there are casualties under an incident is firstly predicted, then in the regression step, samples predicted to have casualties are used to further predict the exact number of the casualties. Using an open-source dataset, this two-step method is validated. The results show that the two-step model performs better than the original regression models. Back propagation(BP) neural network combined with Random Forest performs the best in terms of the death toll and the number of injuries. Among all the two-step models, the lowest mean absolute error (MAE) for the death toll is 1.67 while that for the number of injuries is 4.13, which indicates that this method can accurately predict the number of casualties under emergencies. This study's results are expected to provide support for decision-making on rapid resource allocation and other emergency responses.
A hybrid deep learning-based framework for future terrorist activities modeling and prediction
2022, Egyptian Informatics Journal
Terrorism has led to massive humanitarian and economic crisis due to dire events that affected many countries and caused thousands of deaths and critical damages. In literature, various Artificial Intelligence (AI) based research works have been proposed and enhanced to counter and predict terrorist activities. Practically, Machine Learning (ML) techniques are the most applied. However, with the increasing of the complexity and volume of data, ML algorithms fail to detect and predict accurately terrorist activities. Thus, for understanding the behavior of terrorist actors, we proposed a hybrid Deep Learning (DL) platform based on Convolutional Neural Network (CNN) and Long Short-Term Memory (LSTM) models to learn the temporal features from the Global Terrorism Database (GTD) and to predict future terrorist activities characteristics. The GTD is a well-known database which contains around 190,000 terrorist events and incidents around the world since 1970 until 2020 and incorporates multiple factors, such as the type of weapons used, the attack is successful or not, the kind of attack, and the category of terrorist. First the CNN is used extract complex features of the data, and then these features are forwarded to LSTM model to learn the temporal relationship of the data. Simulation results show that the CNN-LSTM model achieves superior performance for bi-classification tasks which achieves an accuracy more than 96%, while the DNN outperforms the hybrid aforementioned model with accuracy of 99,2% the multi-classification task of predicting terrorist activities. The proposed model shows also a correlation between the occurrence of attacks with the type of weapons used and can accurately predict the type of terrorist attacks with their success rate.
Investigation into the role of human and organizational factors in security work against terrorism at large-scale events
2020, Safety Science
Citation Excerpt :
Therefore, LSE security management personnel should continually monitor the need to upgrade cyber security protection systems and update the training of cyber security professionals. In addition, the cumulative frequency of L5F1 is also 0.33, indicating that terrorism has posed a serious threat to the security and stability of various nations and regions for decades (Meng et al., 2019). China is facing a serious threat of terrorism (Zhang, 2019).
The pervasive human and organizational factors (HOFs) in security work at large-scale events (LSEs) contribute greatly to preventing terrorist attacks. However, as they have not been systematically analyzed, the formulation of relevant precautions could be flawed. This study aims to construct a quantitative model for: (i) systematically analyzing the HOFs in security work against terrorism at LSEs, (ii) predicting the probability of such terrorist attacks, and (iii) diagnosing the most critical HOFs. First, 30 HOFs were systematically identified by modifying the Human Factors Analysis and Classification System (HFACS) and integrating relevant historical data, literature, and expert knowledge. Second, these HOFs were statistically analyzed. Finally, by taking the Beijing 2022 Winter Olympics as an example, a hybrid HFACS-Bayesian Network model was constructed to quantitatively analyze the HOFs based on data collected through questionnaires and expert interviews. The example demonstrates the hybrid model’s capabilities in probability prediction and key factor diagnosis. This study contributes to the establishment of a systematic causation model for analyzing the root causes of the failure of security against terrorism at LSEs, which will enable more holistic incident investigation and more accurate formulation of precautions, as well as helping the development of risk analysis methods in the public security field.
GA-CatBoost-Weight Algorithm for Predicting Casualties in Terrorist Attacks: Addressing Data Imbalance and Enhancing Performance
2024, Mathematics
Attention-based spatial–temporal multi-graph convolutional networks for casualty prediction of terrorist attacks
2023, Complex and Intelligent Systems

View all citing articles on Scopus

Nie Lingyu is currently an undergraduate student in the school of Investigation and Counterterrorism at Chinese People's Public Security University. Her research interest is mainly in the area of public security intelligence. She has participated some scientific research projects and made some achievements.

Song Jiepeng is currently an undergraduate student in the school of Investigation and Counterterrorism at Chinese People's Public Security University. His research interest is mainly in the area of public security intelligence, and he has participated some scientific research projects.

: This paper is for CAEE special section SI-hai. Reviews processed and recommended for publication to the Editor-in-Chief by Associate Editor Dr. G. Ramirez Gonzalez.

View full text

Big data-based prediction of terrorist attacks

Abstract

Introduction

Section snippets

Research methods

Analysis and discussion of experimental results

Conclusion

Conflict of interest

Futur Gener Comput Syst

Introducing the global terrorism database

Terror Political Violence

Terrorism and India: an economic perspective

Qual Quant

Domestic versus transnational terrorism: data, decomposition, and dynamics

J Peace Res

Visualizing patterns in a global terrorism incident database

Environ Plann B Plan Des

Naive Bayesian classifier

Artificial neural networks

IEEE Circuits Devices Mag

Scaling up the accuracy of Naive-Bayes classifiers: a decision-tree hybrid

An improved ID3 decision tree algorithm

Detecting terrorism incidence type from news summary

Attack type prediction using hybrid classifier

Majority vote and decision template based ensemble classifiers trained on event related potentials for early diagnosis of Alzheimer's Disease

Int J Comput Appl