AI-Supported Smart Search for Multi-Period Historical Maps Using the mapKurator Place Name Index

Abstract

Placenames, which link maps to historical records are vital for understanding spatiotemporal societal changes. Traditional workflows require scanned historical maps to be georeferenced before publishing in GIS-readable formats or web map services. Querying, however, typically remains limited to layer metadata, hindering historical placename identification across periods. This paper presents a complete pipeline that integrates batched placename extraction and large language models (LLMs) to enable smart spatial search. Using mapKurator, we batch-extracted text from 20th-century Web Map Tile Service format maps with simultaneous geolocation capture. Indexed placenames were stored in an Elasticsearch database integrated with OpenAI’s LLMs to extract spatiotemporal information from free-text user queries. An elastic, fuzzy-search engine retrieved relevant results exportable for GIS applications. By merging LLMs with mapKurator, a smart spatial search system was developed that efficiently compiles, visualizes, and overlays historical map layers on a Web GIS platform, significantly enhancing the searchability and analysis of historical maps.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 2 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.