This project demonstrates a GeoAI pipeline for extracting addresses from unstructured text and transforming them into structured geospatial data using geoparsing and geocoding techniques.
The solution combines:
- Natural Language Processing (NLP)
- Quantum-inspired NLP (QNLP / QLSTM)
- Geoparsing & address normalization
- Geocoding for GIS-ready outputs
The project is designed as a research-to-engineering showcase, illustrating how advanced NLP models can be applied to real-world geospatial analytics and decision-support systems.
-
Address detection in unstructured text
(incident reports, messages, descriptions, logs) -
Geoparsing & normalization
Extraction of location entities (streets, buildings, settlements) -
Geocoding to spatial coordinates
Conversion of textual addresses into GIS-compatible formats (points / attributes) -
Hybrid NLP models
Including QNLP (Quantum Natural Language Processing) and classical deep learning models -
GIS-ready outputs
Structured tables suitable for ingestion into ArcGIS / PostGIS / GeoPandas pipelines
- Emergency and incident reports analysis
- Infrastructure maintenance logs
- Utility service requests & complaints
- Crisis response and situational awareness
- Preprocessing textual data for GIS systems
- Python
- NLP: spaCy, NLTK
- Deep Learning: LSTM / QLSTM
- Quantum NLP: Qiskit, PennyLane (research layer)
- Geospatial: GeoPandas, Shapely (integration-ready)
- Notebooks: Jupyter
This repository demonstrates models, methods, and workflows using public or synthetic examples.
In production scenarios (e.g. critical infrastructure, utilities, emergency systems):
- Real datasets and integrations are maintained in private repositories
- Full pipelines can be demonstrated using anonymized data upon request
- GeoAI & Spatial Intelligence
- Geoparsing & Geocoding
- Unstructured-to-structured data pipelines
- AI-assisted GIS preprocessing
- Research-to-production AI workflows
Tetiana Starovoit
Senior GeoAI & Geospatial Software Engineer
PhD Candidate (GeoAI)
GitHub: https://github.com/Tania526-sudo
LinkedIn: https://www.linkedin.com/in/tetiana-starovoit-61246200/
This project is intended for demonstration and research purposes and does not expose sensitive operational data.