MappingChange: A Temporal and Semantic Knowledge Base of the Scottish Gazetteers (1803–1901)

MappingChange is a reproducible resource that builds a temporal and semantic knowledge base from ten editions of the Gazetteers of Scotland (1803–1901), digitized by the National Library of Scotland. It includes structured DataFrames, an RDF knowledge graph, enrichment scripts, and interactive Jupyter notebooks to support historical and spatial analysis of place-based descriptions.

The pipeline extracts over 50,000 entries using GPT-4, enriches them via named entity recognition (NER), semantic clustering, and georesolution, and models the results using the tage Textual Ontology (HTO) The resulting knowledge base comprises:

1. Article-level place descriptions and metadata,

2. Cross-edition conceptual alignments using sentence embeddings and external links (Wikidata, DBpedia),

3. Geospatial annotations with temporal and spatial modeling (CRMgeo, GeoSPARQL).

The resource supports research in digital humanities, linked data, and historical NLP. All data and code are released under open licenses and are integrated into the Frances semantic platform for exploration and reuse. GitHub repository MappingChange