To support (a large amount of) company names and other proper names that consist of multiple tokens, I would like to 'spell check' those word groups. Since a multi-word checker spell checker is not in planning, I as thinking of realizing the following:
Add each component of such a group (e.g. Coca Cola) as a separate word in the speller and postag dictionary as (e.g.)
Coca:POW:001:001 (part of word group: #1: part 1)
Cola:POW:001:002 (part of word group: #1: part 2)
Then have a rule warning for the loose parts of this group
(postag_regexp="yes" postag="POW.*")
And use antipatterns to allow for groups that belong together and are in the right order
(<antipattern><token postag="POW:001:001"/><token postag="POW:001:002"/></antipattern>)
But to make that flexible, I would need the word group number of be identical.
Is there a way to do that?