Deterministic canonical stem of an entity surface: casefold, collapse
whitespace/underscores, strip surrounding markdown/quotes, drop a trailing
period, then strip ONE leading article (the/a/an) and ONE leading possessive
determiner (my/your/his/her/its/our/their), plus a trailing possessive 's.
This is the dedup key: a name's spacing variants, short forms, brand-only
surfaces, and possessive surfaces all collapse toward one stem so variants of
the same thing unify.
Deterministic canonical stem of an entity surface: casefold, collapse whitespace/underscores, strip surrounding markdown/quotes, drop a trailing period, then strip ONE leading article (the/a/an) and ONE leading possessive determiner (my/your/his/her/its/our/their), plus a trailing possessive
's.This is the dedup key: a name's spacing variants, short forms, brand-only surfaces, and possessive surfaces all collapse toward one stem so variants of the same thing unify.