How to detect "Proper Nouns" ?

How about adding the tag “Proper” to the Noun‘s tagset ?

In my opinion it could be for the Nouns Uncountable | Organization | Name | Person | Place

but perhaps should also be Acronym

and for Uncountable only if it follows a Determiner or Possessive

[spencermountain]

hey Gagan, yeah it’s a little hard to determine. If we understand a proper noun as a thing that should-be-capitalized, the best way to find this is probably doc.match('(#Person|#Place|#Organization)').

the not-in-dictionary strategy will grab a lot of miscellaneous things, which I wouldn’t recommend.

There will also be some false-positives with (#Person|Place|#Organization) - words like ‘brother’ or ‘deputy editor’ may slip in, same for ‘south Main St.’ and so on.

[spencermountain]

hey, I’ve added support for #ProperNoun tag in 11.5.0. It basically just runs-through the people|places|orgs after the tagger has run, and tags things as a ProperNoun.

it gets a little harder when you change the word, the tag should disappear. i think i’ve mostly guarded against this, but who really knows. lemme know if you see anything funny. thanks.