PEPAdb MVP: Advanced Data Querying, GIS Filtering, and Machine Learning Inference for Archaeological Materials

PEPAdb (https:pepadb.us.es) is an existing research-oriented database for archaeological materials, currently offering basic tabular queries and map visualisation. The goal of this project is to develop a Minimum Viable Product that significantly extends the platform’s functionality by enabling advanced data querying, integrated spatial filtering, and automated machine-learning–based analysis within a single web interface.

The project will integrate a PostgreSQL/PostGIS database, GeoServer-based spatial services, and pre-existing machine learning models deployed as containerised microservices. Researchers will be able to explore archaeological datasets using multi-parameter filters (e.g. chronology, material, provenance, spatial location), visualise results on interactive maps, upload their own analytical or spectral data, and obtain automated predictions (e.g. geological provenance). The project is research-driven but technically well-scoped, offering students experience with modern web development, APIs, GIS, and AI/ML integration.

The MVP will deliver a functional web-based research interface integrated into the existing PEPAdb platform. Users will be able to: - Query individual archaeological records using structured filters (material, chronology, site, provenance) - Apply spatial filters through an interactive map linked to the database - Upload compositional or spectral datasets - Run a containerised machine-learning model (VORTEX) to obtain provenance predictions - Visualise uploaded spectra alongside reference spectra from the database - Download filtered datasets and prediction results (CSV / GeoJSON)