NextStepAstro birthplace dataset — GeoNames

Source: https://download.geonames.org/export/dump/
GeoNames: https://www.geonames.org/
License: Creative Commons Attribution 4.0 International
https://creativecommons.org/licenses/by/4.0/

Snapshot retrieved: 2026-09-20
Asset version: geonames-2026-09-20-v2
Source files: allCountries.zip, countryInfo.txt, admin1CodesASCII.txt,
admin2Codes.txt. Exact source SHA-256 hashes are in the snapshot manifest.json.

Modified by NextStepAstro: retain current populated places (feature class P),
excluding historical capitals, historical, abandoned, and destroyed places
(PPLCH/PPLH/PPLQ/PPLW). No population threshold is applied. Reject invalid IDs,
coordinates, and missing country labels. Preserve names, useful aliases,
country/admin context and original numeric coordinates. Normalize search names
with lowercase, Unicode NFD, removal of nonspacing marks, and word separators.
Omit elevation, timezone fields/offsets and unrelated source attributes.

Partition by normalized name/alias prefix, starting at two Unicode characters;
large branches subdivide, with ranked previews at shorter prefixes. Branches are
packed into 2,048 deterministic FNV-1a hash buckets (UTF-8 prefix modulo 2048).
Schema v2 deduplicates country/admin labels into a shared contexts.bin table,
stores one record per ID/bucket in column arrays, reconstructs primary-name search
aliases, and delta-encodes node references. Places, useful aliases and numeric
coordinates are preserved; the optimization introduces no additional place filters.
Buckets are gzip-compressed static .bin assets. The application searches decoded data in the
browser; this is not the GeoNames hosted API. No OSM data is used.

Data is provided as is, without warranty of accuracy, timeliness, or completeness.
Coordinates represent places, not verified hospitals/addresses. Missing places
can be entered manually without invented coordinates. No endorsement by GeoNames
is implied. Source data and attribution remain available to downstream users.