Place Names API

Auto-suggests a list of place names based on a supplied prefix.

Data generation

First run the init-db docker compose service: this will download placename data from Gov.UK's ONS GeoPortal and unpack it into a SQLite3 database (into the ./data folder).

A list of placenames was extracted from the sqlite db using the following query:

SELECT DISTINCT place23nm FROM place_names

This was then uploaded to ChatGPT along with the following prompt:

Here's a list of over 62K placenames from around the UK. I want you to read the entire file, and produce a CSV of exactly the same length: field 1 should be the placename from the text file; field 2 should be a relevancy score ranging from 0.0 to 1.0.

You should determine the relevancy based on your knowledge of the given place, how popular it is, or how likely someone would want to find out information about that place. So for example "Edinburgh" would score much higher than "Edinbane". Likewise, it is likely that someone would be more interested in "Nottingham" rather than "Nottinghamshire".

Make sure you give a relevancy score for every line. Just do it all in one go - I dont want a sample of 50 or anything. Give me a link where i can download the CSV.

This results in the gzipped CSV file ./data/placenames_with_relevancy.csv.gz which is then loaded into a trie structure when the server starts.