Adjective Target Resolution
|Submitter Email:||click here to access email|
Does anyone know of an algorithm to determine if a term is modified by a
particular adjective, or a likely place to look for such?
Ideally, it would be something that could be feasibly run on all adjectives
in a large corpus.
Since I have had little success in finding such a thing, I may need to
develop one. I've begun to work towards this end using descriptions of
adjective usage found in grammar books. These tend to be targetted at
people learning or writing the language, I would appreciate references to
any materials that you might think are appropriate to this task.
While musing on this, I've come into some questions:
Adjectives used attributively modify nouns directly, usually by preceding
them. Is it reasonable to assume then that an adjective is modifying the
rest of a noun phrase containing it?
Predicative adjectives are the object of a copular verb. In English, does
it simplify the problem of resolving the subject to only consider such
verbs? For example, in this case, would the subject be guaranteed to
preceed the verb where it wouldn't be in general (aside from questions like
''How sunny is it?'') Could I then, eliminating clauses bracketed by
commas, presume with some accuracy to say the noun phrase preceding a link
verb is the subject of said? Any good references on this topic? I know
some parsers provide such resolution but these tended not to be performant.
Beyond normal attributive and predicative use, what are some notable cases
where special treatment would be required to find the target of an adjective?
I appreciate whatever input you may be able to give and will of course post
a summary of responses.
Sums main page