Files

Efficient Named-entity Recognition for compromise
npm install compromise-wikipedia

This experimental plugin provides a highly-compressed list of ~38 thousand popular wikipedia articles, and efficiently scans a text for them.

The plugin is approximately 300kb, minified. It does not include wikipedia's redirects (which it should!).

It is intended as a proof-of-concept, for compressing a large lexicon onto the client-side.

Wikipedia

import plg from 'compromise-wikipedia'
nlp.extend(plg)

let doc = nlp('You could always go to McGill, the Harvard of Canada!')
let m = doc.wikipedia()
m.json()
// [...]

MIT