NLP and LLMs applied to the geospatial domain

Versão em português

Overview¶

This research strand connects natural language, conceptual models and geospatial practice, using natural language processing, voice recognition and large language models.

Why it matters¶

Heterogeneous geospatial data and specialized workflows often face semantic, terminological and operational barriers. This strand investigates how language-based methods can support interpretation, validation and integration.

Goals¶

• explore NLP use cases in geospatial contexts;

• test LLM applications for geospatial reasoning and practice;

• support semantic alignment and validation of heterogeneous structures;

• bridge natural language and specialized geospatial workflows.

Associated output¶

• Mapping with Words: Integrating Large Language Models into Geospatial Practice

• Reasoning in a Large Language Model (LLM) for validating tags used in OpenStreetMap

• Potential applications of natural language processing for geodetic sciences

• Natural Language Processing and Voice Recognition for Geolocation and Geospatial Visualization in Notebook Environment