Subrahmanian Handbook of Computational Approaches to Counterterrorism

2013
ISBN: 978-1-4614-5311-6
Verlag: Springer
Format: PDF
Kopierschutz: 1 - PDF Watermark

E-Book, Englisch, 578 Seiten

Terrorist groups throughout the world have been studied primarily through the use of social science methods. However, major advances in IT during the past decade have led to significant new ways of studying terrorist groups, making forecasts, learning models of their behaviour, and shaping policies about their behaviour. Handbook of Computational Approaches to Counterterrorism provides the first in-depth look at how advanced mathematics and modern computing technology is shaping the study of terrorist groups. This book includes contributions from world experts in the field, and presents extensive information on terrorism data sets, new ways of building such data sets in real-time using text analytics, introduces the mathematics and computational approaches to understand terror group behaviour, analyzes terror networks, forecasts terror group behaviour, and shapes policies against terrorist groups. Auxiliary information will be posted on the book's website. This book targets defence analysts, counter terror analysts, computer scientists, mathematicians, political scientists, psychologists, and researchers from the wide variety of fields engaged in counter-terrorism research. Advanced-level students in computer science, mathematics and social sciences will also find this book useful.

Autoren/Hrsg.

Subrahmanian, V.S.

Subrahmanian, V. S.

Inhaltsverzeichnis

1;Preface;6
2;Acknowledgements;14
3;Contents;16
4;Part I Data and Data Acquisition;20
4.1;The Global Terrorism Database, 1970 –2010;21
4.1.1;1 Introduction;21
4.1.1.1;1.1 Terrorism Data from Open Sources;23
4.1.1.1.1;1.1.1 Limitations of Event Databases;25
4.1.1.1.2;1.1.2 Strengths of Event Databases;26
4.1.1.2;1.2 World-Wide Terrorism
;27
4.1.2;2 Conclusions;37
4.1.3;A.1 Appendix A Countries Listed Under Each Region According to GTD;38
4.1.4;References;40
4.2;Automated Coding of Political Event Data;41
4.2.1;1 Introduction and Overview;41
4.2.1.1;1.1 Human Versus Machine Coding;43
4.2.2;2 Text Acquisition and Formatting;46
4.2.2.1;2.1 Filtering: Irrelevant Stories;47
4.2.2.2;2.2 Filtering: Duplicates;48
4.2.3;3 Coding Ontologies;49
4.2.3.1;3.1 Events;50
4.2.3.2;3.2 Actors;53
4.2.4;4 Actor Dictionaries and Named Entity Recognition;55
4.2.5;5 Pre-processing Using NLP Tools;56
4.2.6;6 Coding and Post-processing;60
4.2.6.1;6.1 Cluster Processing;60
4.2.6.2;6.2 One-A-Day Filtering;61
4.2.6.3;6.3 Sophisticated Error Detection/Correction;61
4.2.7;7 Open Issues;62
4.2.7.1;7.1 Geolocation;62
4.2.7.2;7.2 Machine Translation;62
4.2.7.3;7.3 Real-Time Coding;63
4.2.8;8 Conclusion;65
4.2.9;References;66
4.3;Automatic Extraction of Events from Open Source Text for Predictive Forecasting;68
4.3.1;1 Introduction;68
4.3.2;2 Task Description;70
4.3.3;3 System Descriptions;70
4.3.3.1;3.1 Tabari;70
4.3.3.2;3.2 BBN SERIF;71
4.3.4;4 Experiment Design;75
4.3.4.1;4.1 Evaluation Corpus;75
4.3.4.2;4.2 Evaluation Procedure;75
4.3.5;5 Evaluation Results;76
4.3.5.1;5.1 Overview;76
4.3.5.2;5.2 Comparison to Previous Studies;77
4.3.5.3;5.3 Error Analysis;78
4.3.5.4;5.4 System Overlap;80
4.3.5.5;5.5 Historical Events;80
4.3.5.6;5.6 Topic Filtering;81
4.3.5.7;5.7 Adapting to New Corpora;82
4.3.6;6 Conclusion;83
4.3.7;References;83
4.4;Automated Coding of Decision Support Variables;85
4.4.1;1 Introduction;85
4.4.2;2 Related Work;86
4.4.3;3 Automatic Coding Engine;87
4.4.3.1;3.1 Preprocessing;89
4.4.3.2;3.2 Linguistic Sensors;90
4.4.3.3;3.3 Logic Layer;91
4.4.4;4 Implementation and Experiments;93
4.4.4.1;4.1 Precision and Recall;94
4.4.4.2;4.2 Time;95
4.4.5;5 Conclusions and Future Work;95
4.4.6;References;96
5;Part II Behavioral Models and Forecasting;97
5.1;Qualitative Analysis & Computational Techniques for the Counter-Terror Analyst;98
5.1.1;1 Introduction;98
5.1.1.1;1.1 Counter-Terror Research Needs
;98
5.1.1.2;1.2 Qualitative Research Overview;99
5.1.1.2.1;1.2.1 Contrasting Qualitative and Quantitative Research;100
5.1.1.2.2;1.2.2 Qualitative vs. Quantitative Research in the Context of Counter-Terrorism;100
5.1.1.3;1.3 Understanding Terrorist Group Behavior;101
5.1.1.3.1;1.3.1 Employing the Strategic Perspective;101
5.1.1.3.2;1.3.2 Attacking Organizational Weakness;104
5.1.1.3.3;1.3.3 Applications of Communications Theory;106
5.1.1.4;1.4 Studying the Individual Terrorist;107
5.1.1.4.1;1.4.1 Counter-Radicalization Strategies;108
5.1.1.4.2;1.4.2 Facilitating Desertions;109
5.1.2;2 Conclusions;110
5.1.3;References;110
5.2;SOMA: Stochastic Opponent Modeling Agents for Forecasting Violent Behavior;113
5.2.1;1 Introduction;113
5.2.2;2 Representing Terror Group Behavior: Action Probabilistic Logic Programs;115
5.2.3;3 Forecasting Terror Group Behavior: Finding the Most Probable World;121
5.2.3.1;3.1 A First Approach to Forecasting in SOMA;123
5.2.3.2;3.2 Scalable Algorithms for Forecasting Terror Group Behavior;124
5.2.3.2.1;3.2.1 Head-Oriented Processing;125
5.2.3.2.2;3.2.2 Randomized Heuristic Behavioral Forecasts;130
5.2.4;4 Distributed Computation for Forecasting in SOMA;131
5.2.4.1;4.1 Parallelism for Reducing Computation Time;131
5.2.4.2;4.2 Parallelism for Increasing Computational Capacity;132
5.2.4.3;4.3 Parallelism for Improving Solution Accuracy;134
5.2.5;5 Applications of ap-Programs;136
5.2.6;6 Conclusions;139
5.2.7;References;140
5.3;Data-based Computational Approaches to ForecastingPolitical Violence;142
5.3.1;1 Introduction and Overview;142
5.3.1.1;1.1 The Development of Technical Political Forecasting;144
5.3.2;2 Data Sources;145
5.3.2.1;2.1 Structural Data;146
5.3.2.2;2.2 Dyadic Data;147
5.3.2.3;2.3 Atomic Event Data;147
5.3.2.4;2.4 Composite Event Data;148
5.3.2.5;2.5 Social Media and Other Unstructured Data Sources;148
5.3.2.6;2.6 The Challenges of Data Aggregation;149
5.3.2.6.1;2.6.1 Actors;149
5.3.2.6.2;2.6.2 Actions;150
5.3.2.6.3;2.6.3 Temporal;150
5.3.3;3 Statistical Approaches;150
5.3.3.1;3.1 Cross-Sectional Regression and Logit;151
5.3.3.2;3.2 Classical Time Series;152
5.3.3.3;3.3 Vector Autoregression Models;154
5.3.3.4;3.4 Event-History and Survival Models;155
5.3.3.5;3.5 Rare-Events Models;156
5.3.4;4 Algorithmic Approaches;158
5.3.4.1;4.1 Supervised Cross-Sectional Classification Methods;159
5.3.4.1.1;4.1.1 Linear Approaches;159
5.3.4.1.2;4.1.2 Neural Networks;159
5.3.4.1.3;4.1.3 Tree-Based Algorithms;160
5.3.4.2;4.2 Unsupervised Methods;161
5.3.4.2.1;4.2.1 Dimension Reduction;161
5.3.4.2.2;4.2.2 Clustering;161
5.3.4.3;4.3 Sequence Development: Hidden Markov Models;162
5.3.4.4;4.4 Sequence Analysis: Sequence Matching;163
5.3.4.4.1;4.4.1 Archetypal Sequence Matching;164
5.3.4.4.2;4.4.2 Convex Algorithms;165
5.3.5;5 Network Models;166
5.3.5.1;5.1 Social Network Analysis Models;166
5.3.5.2;5.2 Geo-spatial Models;167
5.3.6;6 Conclusion;167
5.3.7;References;169
5.4;Using Hidden Markov Models to Predict Terror Before it Hits (Again);176
5.4.1;1 Introduction;176
5.4.1.1;1.1 Hidden Markov Models;177
5.4.1.2;1.2 Issues and Implications;179
5.4.1.3;1.3 Data Development and Pre-processing;179
5.4.1.4;1.4 Training (Baum-Welch estimates);182
5.4.1.4.1;1.4.1 Approach, Initial Estimates and Alternative Models;182
5.4.1.4.2;1.4.2 Sequence Length, Iterations and Losses;183
5.4.1.4.3;1.4.3 Global and Prior Estimates;183
5.4.1.4.4;1.4.4 Global, Cut (Training Set) and Most Recent Densities;184
5.4.1.4.5;1.4.5 Optimization (Viterbi State Trajectories);184
5.4.2;2 Forecasting;185
5.4.2.1;2.1 Iraq and Afghanistan Results;186
5.4.2.2;2.2 Testing of Results and Technical Discussion;189
5.4.3;3 Conclusions;190
5.4.4;4 Training;191
5.4.5;A.1 Appendix A: Technical Details;191
5.4.6;References;192
5.5;Forecasting Group-Level Actions Using Similarity Measures;194
5.5.1;1 Introduction;194
5.5.1.1;1.1 Related Work;195
5.5.1.2;1.2 Contributions and Organization of This Work;197
5.5.2;2 Behavioral Time Series Data;197
5.5.3;3 A Formal Vector Model of Agent Behaviors;198
5.5.4;4 Algorithms for Forecasting Agent Behavior;199
5.5.4.1;4.1 Distance Functions;199
5.5.4.2;4.2 The CONVEXk_NN Algorithm;201
5.5.4.3;4.3 The CONVEXMerge Algorithm;203
5.5.5;5 Implementation and Experiments;205
5.5.6;6 Forecasting Situations;208
5.5.7;7 Conclusions;210
5.5.8;References;211
5.6;Forecasting the Use of Violence by Ethno–Political Organizations: Middle Eastern Minorities and the Choice of Violence;213
5.6.1;1 Introduction;213
5.6.2;2 Efforts at Forecasting in Past;214
5.6.3;3 Forecasting Ethnic Violence: MAROB;215
5.6.4;4 Forecasting from Engineering to the Social Sciences;216
5.6.5;5 Probabilistic Modeling Process Overview;219
5.6.5.1;5.1 Imputation of Missing Values;220
5.6.5.2;5.2 Factor Selection;220
5.6.5.3;5.3 Massage Data;222
5.6.5.4;5.4 Classification;223
5.6.5.5;5.5 Validation and Performance Assessment;226
5.6.6;6 Sensitivity Analysis;227
5.6.7;7 Classification and Forecasting Results;228
5.6.8;8 Conclusion;231
5.6.9;Appendix;232
5.6.10;References;234
5.7;Forecasting Changes in Terror Group Behavior;237
5.7.1;1 Introduction;237
5.7.2;2 CAPE Architecture;238
5.7.2.1;2.1 SitCAST Situation Forecaster;241
5.7.2.2;2.2 SitCAST and CONVEX;242
5.7.2.3;2.3 The CAPE Algorithms;245
5.7.2.3.1;2.3.1 The Change Table;245
5.7.2.4;2.4 Learning Change Indicators from the Change Table;247
5.7.2.5;2.5 The CAPE-Forecast Algorithm;250
5.7.3;3 Implementation and Experiments;251
5.7.4;4 Related Work;252
5.7.5;5 Conclusions;254
5.7.6;References;255
5.8;Using Temporal Probabilistic Rules to Learn Group Behavior;256
5.8.1;1 Introduction;256
5.8.2;2 Modeling Group Behavior with Temporal Probabilistic Logic Programs;258
5.8.2.1;2.1 Database Schema for a Group's Past Behavior;258
5.8.2.2;2.2 Syntax;259
5.8.3;3 Automatically Learning Rules from Historical Data;262
5.8.3.1;3.1 Automatic Extraction of TP-Rules;262
5.8.3.1.1;3.1.1 SOMA Rules;262
5.8.3.1.2;3.1.2 Subrahmanian-Ernst Algorithm: Preliminaries;263
5.8.3.1.3;3.1.3 The Subrahmanian-Ernst Algorithm and an Adaptation to TPLPs;266
5.8.3.2;3.2 Toward Converting TP-Rules into Policy Recommendations;268
5.8.3.2.1;3.2.1 Computational Policies;269
5.8.3.2.2;3.2.2 Iteratively Computing All Policies;270
5.8.4;4 Policy Recommendations and Lashkar-e-Taiba;272
5.8.4.1;4.1 Experimental Methodology and Learned Rules;272
5.8.4.2;4.2 Policies That Potentially Eliminate or Reduce Violent Attacks by Lashkar-e-Taiba;274
5.8.5;5 Conclusions and Directions for Future Research;275
5.8.6;References;276
6;Part III Terrorist Network Analysis;278
6.1;Leaderless Covert Networks: A Quantitative Approach;279
6.1.1;1 Introduction;279
6.1.2;2 Covert Network Models and Centrality;281
6.1.3;3 Homogeneous Networks;282
6.1.4;4 Heterogeneous Networks;285
6.1.5;5 Case: Jemaah Islamiyah's Bali Bombing;287
6.1.6;6 Conclusion;288
6.1.7;7 Methods Summary;289
6.1.7.1;7.1 Information Measure I;289
6.1.7.2;7.2 Homogeneous Secrecy Measure Shom;289
6.1.7.3;7.3 Heterogeneous Secrecy Measure Shet;290
6.1.7.4;7.4 Balanced Trade-Off Performance Measure µ;290
6.1.7.5;7.5 Game Theoretic Centrality;290
6.1.8;References;291
6.2;Link Prediction in Highly Fractional Data Sets;293
6.2.1;1 Introduction;293
6.2.2;2 Background;295
6.2.2.1;2.1 Social Networks of Terrorists;295
6.2.2.2;2.2 Link Prediction;295
6.2.3;3 Social Network Datasets;296
6.2.4;4 Methods and Experiments;301
6.2.4.1;4.1 Experimental Setup;301
6.2.4.2;4.2 Feature Extraction;303
6.2.5;5 Results;304
6.2.6;6 Conclusion;308
6.2.7;References;308
6.3;Data Analysis Based Construction and Evolution of Terrorist and Criminal Networks;311
6.3.1;1 Introduction;311
6.3.2;2 Network Construction;313
6.3.2.1;2.1 Network Re-construction;316
6.3.3;3 Network Partitioning;319
6.3.3.1;3.1 Method;321
6.3.3.1.1;3.1.1 Construction;321
6.3.3.1.2;3.1.2 Partition;322
6.3.3.1.3;3.1.3 Computation;323
6.3.3.2;3.2 Results;323
6.3.4;4 Link Prediction;325
6.3.4.1;4.1 Link Prediction Method;326
6.3.4.2;4.2 Results and Discussions;329
6.3.4.2.1;4.2.1 Success Criteria;329
6.3.5;5 Conclusions;330
6.3.6;References;330
6.4;CrimeFighter Investigator: Criminal Network Sense-Making;332
6.4.1;1 Introduction;332
6.4.2;2 Criminal Network Sense-Making;333
6.4.2.1;2.1 Criminal Network Investigation Model;335
6.4.2.2;2.2 Sense-Making Tasks;336
6.4.3;3 CrimeFighter Investigator;342
6.4.3.1;3.1 Conceptual Model;344
6.4.3.2;3.2 Computational Model;345
6.4.3.3;3.3 Structural Parser;351
6.4.4;4 Scenario: Investigating Linkage Between DNRI and AQAM;354
6.4.4.1;4.1 The Scenario;355
6.4.4.2;4.2 Summary;360
6.4.5;5 Related Work;361
6.4.6;6 Conclusion and Future Work;364
6.4.7;References;365
7;Part IV Systems, Frameworks, and Case Studies;369
7.1;The NOEM: A Tool for Understanding/Exploring the Complexities of Today's Operational Environment;370
7.1.1;1 Introduction;370
7.1.1.1;1.1 A Step Forward;373
7.1.1.2;1.2 Supporting Stability Operations;374
7.1.1.2.1;1.2.1 Modeling and Simulation Support to Stability Ops;376
7.1.1.3;1.3 The National Operational Environment Model;382
7.1.2;2 NOEM Overview;385
7.1.2.1;2.1 The Model;385
7.1.3;3 Using the NOEM Tools;388
7.1.3.1;3.1 Point or Event Based Analysis;388
7.1.3.2;3.2 Prospective Analysis
;391
7.1.3.3;3.3 Model Validation
;397
7.1.3.3.1;3.3.1 Verification and Face Validation;398
7.1.3.3.2;3.3.2 Inverse V&V;401
7.1.4;4 Conclusion;404
7.1.5;5 Disclaimer;404
7.1.6;References;404
7.2;A Multi-Method Approach for Near Real Time Conflict and Crisis Early Warning;407
7.2.1;1 Introduction;407
7.2.1.1;1.1 Building on Previous Research;407
7.2.1.2;1.2 DARPA's ICEWS Program;410
7.2.1.3;1.3 Adjusting to Operational Reality: Lessons Learned from the ICEWS Program;412
7.2.2;2 Components of the ICEWS System;414
7.2.2.1;2.1 iTRACE;414
7.2.2.2;2.2 iSENT;416
7.2.2.3;2.3 iCAST;419
7.2.3;3 Summary and Conclusion;422
7.2.4;References;423
7.3;A Realistic Framework for Counter-terrorism in Multimedia;425
7.3.1;1 Introduction;425
7.3.2;2 Violence in Videos;427
7.3.2.1;2.1 Violence Identification in Videos;427
7.3.2.2;2.2 Semantics Extraction in Videos;429
7.3.2.3;2.3 Existing Methods;433
7.3.3;3 Proposed Methodology;435
7.3.3.1;3.1 A Realistic Framework;435
7.3.3.2;3.2 Story-Line of Violent Scene;442
7.3.3.3;3.3 Degree of Violence;443
7.3.4;4 Discussion;443
7.3.5;5 Conclusion;444
7.3.6;References;445
7.4;PROTECT in the Ports of Boston, New York and Beyond: Experiences in Deploying Stackelberg Security Games with Quantal Response;447
7.4.1;1 Introduction;447
7.4.2;2 Background;449
7.4.2.1;2.1 Stackelberg Security Game;449
7.4.2.2;2.2 Deployed Security Applications;450
7.4.3;3 USCG and PROTECT's Goals;451
7.4.4;4 Key Innovations in PROTECT;452
7.4.4.1;4.1 Game Modeling;453
7.4.4.2;4.2 Compact Representation;455
7.4.4.3;4.3 Human Adversary Modeling;457
7.4.5;5 Evaluation;459
7.4.5.1;5.1 Memory and Run-time Analysis;459
7.4.5.2;5.2 Utility Analysis;460
7.4.5.3;5.3 Robustness Analysis;461
7.4.5.4;5.4 USCG Real-World Evaluation;463
7.4.5.5;5.5 Outcomes Following the Boston Implementation;465
7.4.6;6 Lessons Learned: Putting Theory into Practice;465
7.4.7;7 Summary and Related Work;467
7.4.8;References;468
7.5;Government Actions in Terror Environments (GATE): A Methodology that Reveals how Governments Behave toward Terrorists and their Constituencies;470
7.5.1;1 Introduction;470
7.5.2;2 What is known about Government Actions to End Terrorism;471
7.5.3;3 Introducing the GATE Database;476
7.5.3.1;3.1 The Data Described;480
7.5.4;4 Exploring Counterterrorism Effectiveness Using GATE Data;483
7.5.4.1;4.1 The Effectiveness of Israeli Actions on Palestinian Terrorist Violence;485
7.5.4.2;4.2 The Effectiveness of Turkish Actions on Kurdish Terrorist Violence;487
7.5.5;5 Conclusion;488
7.5.6;References;489
8;Part V New Directions;492
8.1;A CAST Case-Study: Assessing Risk in the Niger Delta;493
8.1.1;1 Introduction;493
8.1.1.1;1.1 Component 1: Theory;494
8.1.1.2;1.2 Component 2: Data;497
8.1.1.2.1;1.2.1 Context;497
8.1.1.2.2;1.2.2 Events;498
8.1.1.2.3;1.2.3 Participatory Early Warning and Conflict Mapping;501
8.1.1.3;1.3 Component 3: Analysis;505
8.1.1.3.1;1.3.1 Background: The Origins of Conflict in the Niger Delta;505
8.1.1.3.2;1.3.2 The Niger Delta in 2011;512
8.1.1.3.3;1.3.3 Conclusion and Outlook for the Future: Mitigating Terrorism Risks;515
8.1.2;References;516
8.2;Policy Analytics Generation Using Action Probabilistic Logic Programs;518
8.2.1;1 Introduction;518
8.2.2;2 Preliminaries;520
8.2.2.1;2.1 Syntax;520
8.2.2.2;2.2 Semantics of ap-Programs;521
8.2.3;3 Abductive Queries to Probabilistic Logic Programs;523
8.2.3.1;3.1 Algorithms for BAQA over Threshold Queries;524
8.2.4;4 Cost-Based Abductive Query Answering;528
8.2.5;5 Parallel Solutions for Abductive Query Answering;532
8.2.5.1;5.1 Parallel Selection of Entailing States;533
8.2.5.2;5.2 Parallel Sampling of State Paths;534
8.2.6;6 Experimental Results;535
8.2.6.1;6.1 Empirical Evaluation of Algorithms for CBQA;535
8.2.6.2;6.2 Empirical Evaluation of Parallel Algorithms for CBQA;538
8.2.7;7 Related Work;542
8.2.8;8 Conclusions;543
8.2.9;References;544
8.3;The Application of Search Games to Counter Terrorism Studies;546
8.3.1;1 The Mathematics of Search Games;548
8.3.1.1;1.1 A Brief History of Search Games;548
8.3.1.2;1.2 Search Games on Networks;550
8.3.1.3;1.3 Games of Degree and Multi-agent Games;552
8.3.2;2 Some Counter-Terrorism Search Games;553
8.3.2.1;2.1 The Patrolling Game;553
8.3.2.2;2.2 Disperse or Unite;555
8.3.2.3;2.3 Finding a Moving Fugitive;556
8.3.2.4;2.4 Some Remarks on Multi-agent Search Games;558
8.3.3;3 Summary;558
8.3.4;References;559
8.4;Temporal and Spatial Analyses for Large-Scale Cyber Attacks;561
8.4.1;1 Introduction;561
8.4.2;2 Intrusion Detection and Alert Correlation;562
8.4.3;3 Attack Characterization and Prediction;564
8.4.4;4 Host Clustering and Botnet Detection;567
8.4.5;5 Coordinated Attacks;568
8.4.6;6 Spatial and Temporal Analyses for Coordinated Attacks;571
8.4.7;7 Conclusion;578
8.4.8;References;578