Air quality research depends on diverse data sources, from government monitoring stations to community sensors and weather forecasts. Finding and describing these datasets is often messy and time-consuming. The goal of the project is to explore how Large Language Models (LLMs) can act as a research assistant to automatically discover datasets from open portals and generate useful metadata about pollutants, units, frequency, and coverage.
The project will design and implement an LLM-powered application that helps researchers work with urban air quality data. The system will be able to discover datasets from open data portals, analyze them (e.g., compare air quality levels across different parts of a city or region, track changes over time), and visualize results (such as maps, time series, or dashboards). A key feature will be the incorporation of geospatial metadata to enable mapping and location-aware analysis. The goal is to prototype an interactive assistant that not only automates dataset discovery and metadata generation, but also supports meaningful analysis and communication of air quality data in an accessible and spatially explicit way.
As an extension, the project can also explore ethical and legal considerations, including licensing of open datasets and the responsible use of AI in research support systems.