Landnotes

Landnotes shows Wikipedia places and events on top of OpenStreetMap - try it out!

There are currently 6.5 million events extracted from 400,000 articles using Google Gemini. If anything doesn't make sense, blame it on the AI! There might be 10 times more events to extract by scanning all of Wikipedia. Learn more about future directions in the project's documentation.

Stack

Landnotes is built on top of the following technologies :pray:

Web application:

• OpenStreetMap for the maps, via LeafletJS
• Svelte 5 for the components and interactions
• Cloudflare (D1, R2, Workers) for the infrastructure.
• And of course Wikipedia and its API

Data curation:

• Google Gemini for the event extraction
• wiki-dump-extractor (written specially for this project) for handling, indexing, parsing the wikipedia page dump.
• mwparserfromhell for parsing Wikipedia markup into plain text.

Development

Run the website locally with:

cd website
npm run dev:remote-dbs

Or to run with local databases:

cd website
npm run dev

However for the local-database mode to run properly you need a local database worker (from Cloudflare):

• First build a database using the scripts in data_curation (that's the hard part! Working on providing a test database).
• Then install wrangler on your machine.

You can run the database worker with

cd worker
wrangler dev