CURRENT RESEARCH • AI • SPATIAL DATA • DIGITAL TWIN
Synthetic Spatial Crime Data
Building realistic digital environments for a new generation of spatial criminology and experimental crime research.
This project combines generative artificial intelligence, spatial statistics and crime data to develop high-fidelity synthetic datasets that can support theory testing, simulation and evaluation while addressing many of the privacy and access limitations associated with sensitive criminal justice data.
PROJECT OVERVIEW
Building a Digital Laboratory for Crime Research
Crime researchers often depend on administrative datasets that contain sensitive information, vary substantially across jurisdictions and can be difficult to access or share. These limitations make replication, methodological experimentation and large-scale comparative research difficult.
The Synthetic Spatial Crime Data project is developing a digital research infrastructure capable of generating realistic artificial crime incidents and related urban characteristics. The goal is not simply to create artificial records, but to construct synthetic environments that reproduce meaningful spatial, temporal and social relationships found in real cities.
The initial proof of concept focuses on Dallas, Texas, with a longer-term vision of developing scalable research infrastructure capable of supporting comparative research across multiple U.S. metropolitan areas.
PROJECT AT A GLANCE
Synthetic Crime Research Infrastructure
FOCUS
Synthetic spatial and temporal crime data
INITIAL LOCATION
Dallas metropolitan area
CORE METHODS
Generative AI • Spatial modeling • Simulation
WHY SYNTHETIC DATA?
Creating New Opportunities for Experimental Criminology
Well-designed synthetic datasets can create controlled research environments that complement—not replace—real-world criminal justice data.
Privacy
Reduce reliance on identifiable or sensitive incident-level information when developing and testing analytical methods.
Reproducibility
Create research environments that can be shared, repeated and systematically evaluated by other researchers.
Experimentation
Manipulate controlled conditions to evaluate theories, analytical assumptions and alternative crime-prevention scenarios.
Scalability
Develop methods that can eventually support comparative research across multiple cities and metropolitan environments.
DATA FOUNDATION
Modeling the Complexity of Real Cities
Synthetic crime cannot be modeled in isolation. The research environment integrates multiple types of spatial and social information to represent the context in which criminal events occur.
Crime Incident Data
Census & Demographic Data
Human Mobility Data
Foot-Traffic Patterns
Points of Interest
Urban Spatial Infrastructure
RESEARCH APPROACH
From Real Data to Synthetic Environments
The project combines computational methods with spatial criminology to generate, evaluate and ultimately use synthetic urban crime environments.
01
Integrate Data
Combine crime, demographic, mobility and spatial information into a unified research environment.
02
Train Models
Evaluate generative approaches including GANs, variational autoencoders and diffusion models.
03
Generate Crime Data
Produce synthetic incidents designed to reproduce meaningful spatial and temporal crime characteristics.
04
Validate & Simulate
Compare synthetic patterns with empirical data and use validated environments for simulation and experimentation.
CLOUD-ENABLED RESEARCH
Computational Infrastructure for Research at Scale
The proof of concept is being developed with cloud infrastructure and consultation from Amazon Web Services specialists. The computing environment supports large geospatial datasets, GPU-based model training, synthetic data generation and spatial post-processing workflows that would be difficult to conduct on conventional desktop systems.
Research Infrastructure
Data Lake
Large-scale geospatial and criminal justice research data
GPU Training
Machine-learning and generative-model development
Spatial Processing
Geospatial simulation and validation workflows
Generative AI
Advanced synthetic-data experimentation
FUTURE RESEARCH DIRECTION
Toward a Synthetic City
The longer-term vision extends beyond generating individual synthetic crime records. The project seeks to develop an AI-enabled synthetic city in which people, places, movement, social relationships and criminal events can interact within a virtual urban environment.
Such an environment could provide criminologists with a new form of experimental laboratory—one in which researchers can alter conditions, observe emerging crime patterns and test the potential effects of interventions without imposing those experiments on real communities.
Simulate Human Movement
Model Social Behavior
Test Interventions
RESEARCH TEAM
Project Research Team
The project combines expertise in spatial criminology, crime analysis, data science, cybercrime and emerging technologies.
PROJECT RESEARCHER
Dr. Olga B. Semukhina
Department Head & Professor
Department of Criminal Justice
PROJECT RESEARCHER
Dr. Christopher Copeland
Associate Professor
Department of Computer Science & Electrical Engineer
CURRENT STATUS
From Proof of Concept to Research Platform
PHASE 01
Dallas Proof of Concept
Develop and evaluate synthetic spatial crime-data generation using a major metropolitan environment.
PHASE 02
Model Validation
Evaluate whether synthetic datasets reproduce meaningful spatial, temporal and social characteristics of observed crime.
FUTURE
Multi-City Research
Expand the infrastructure to support comparative research, simulation and experimental criminology across multiple cities.
AI • SPATIAL DATA • EXPERIMENTAL CRIMINOLOGY
Interested in Synthetic Crime Data Research?
We welcome interdisciplinary collaboration involving synthetic data, generative AI, geospatial science, criminology, simulation and data-driven public safety research.