pass-in term metadata in lexicon

Try nlp("modulus").nouns().toSingular().out("text") The result is “modulu”, but actually modulus is singular and its plural is moduli Try nlp("moduli").nouns().toSingular().out("text") Result is “moduli” but it should be modulus.

Is there a way to specify to the engine these kind of rare/peculiar words? Like in a set of json document where each document would be structured as

{
    "word":"modulus",
    "type":"noun",
    "plural":"moduli",
    "singular":"modulus"
    // In case word was of type verb then other fields like past-tense etc will be present
}

[spencermountain]

yeah leo, where i was going with your lexicon-as-trie idea was this:

var nlp=require('compromise')
var myLexicon={blah:'whatever'}
let shareableContext = nlp.pack(myLexicon)
...
nlp('say my name say my name, when no one is around you, say baby..', shareableContext)

[spencermountain]

hey guys, this stuff is deployed on v11:

nlp.addWords()
nlp.addTags()
nlp.addRegex()
nlp.addPatterns()
nlp.addPlurals()
nlp.addConjugations()
//or just combine them in:
nlp.plugin()

please check it out, https://github.com/nlp-compromise/compromise/wiki/Plugins https://github.com/nlp-compromise/compromise-plugin cheers!