Research
The research program is organized around a single question: How can geographic evidence, environmental processes, human behavior, urban dynamics, and spatial language be integrated into computational systems that remain interpretable, reproducible, and useful for real decisions?
Core pipeline
Geoinformatics
GIS, remote sensing, spatial databases, geospatial ETL, map algebra, spatial statistics, and reproducible geographic workflows.
ABM + GIS
Integration of agent-based modeling and GIS to explore human behavior, spatial interaction, emergence, mobility, adaptation, and scenario outcomes.
Urban & socio-economic systems
Urban dynamics and socio-economic environments using GIS, spatial analysis, social network analysis, social media data, and computational social science.
Environmental & ecological simulation
Spatially explicit environmental processes, landscape transitions, risk surfaces, cellular automata, system dynamics, and coupled human–environment systems.
WebGeo / WebGIS
Browser-native geospatial computation, spatial APIs, interactive maps, reproducible web workflows, and decision-support interfaces.
GeoNLP / GeoAI / spatial agents
Place-aware NLP, knowledge graphs, graph learning, spatial reasoning, geographic tool use, and agents grounded in maps and environmental evidence.
Research signature
This signature links classical geographic regionalization and landscape thinking with modern simulation, spatial computation, human-behavior modeling, geographic AI, and language-based reasoning.
Areas of expertise
- Integrating agent-based modeling and geographic information systems to explore human behavior in spatially explicit environments.
- Exploring natural and socio-economic environments, particularly urban areas, as coupled spatial systems.
- Combining GIS, spatial analysis, social network analysis, social media data, and agent-based modeling.
- Connecting simulation outputs with WebGIS, GeoAI, GeoNLP, spatial knowledge, and decision-support workflows.
Method stack
Intellectual & methodological influences
The contemporary intellectual and methodological influences emphasized in this profile are V.S. Tikunov, Andrew Crooks, Philippe Caillou, Benoit Gaudou, Arnaud Grignard, Chi Quang Truong, and Patrick Taillandier.
Their methodological influence is reflected in the current emphasis on geoinformatics, agent-based modeling, GIS-integrated simulation, human behavior, urban systems, complex systems, and spatial decision support.
The broader historical lineage begins with Russian natural zonality, landscape, and territorial-regionalization traditions; continues through Chinese natural regionalization and geoinformatics development; and is localized through Vietnamese physical geography, GIS, remote sensing, and applied geoinformatics. The popup separates this historical lineage from contemporary methodological influences and does not present the map as a formal adviser–student genealogy.
Current research position
Department of Engineering and Technology
Vietnam National University Ho Chi Minh City – Campus in Ben Tre, Vinh Long, 930000, Vietnam
The next-generation direction in this profile is concentrated in four linked tracks:
Dynamic landscapes and coupled systems
Landscape dynamics, CA, ABM, system dynamics, ecological-engineering intervention scenarios, and coupled human–environment processes.
Human behavior and urban simulation
Spatially explicit agents, social interaction, mobility, social networks, social media signals, and urban scenario exploration.
Spatial learning and prediction
Spatial machine learning, GNNs, physics-informed models, interpretable prediction, and spatial transfer.
Spatial knowledge and geographic language
Landscape ontologies, knowledge graphs, place-aware corpora, geographic RAG, and map-grounded LLM queries.
Design principles
- Evidence first: every computational output should remain traceable to data, assumptions, and model steps.
- Spatial grounding: language and AI outputs should be connected to explicit locations, scales, and geographic relationships.
- Behavioral realism: agent assumptions and human-decision rules should be explicit, testable, and linked to empirical or contextual evidence.
- Reproducibility: workflows should be executable, inspectable, and portable across desktop and web environments.
- Human-in-the-loop: expert judgment and local knowledge remain part of the modeling system rather than being treated as noise.
- Decision relevance: technical sophistication is useful only when it improves interpretation, comparison, or action.