On 9/13/11 7:44 PM, C.F.Scheidecker Antunes wrote:
I have a question on how to parse anything related to location on a phrase.
If you say: "The party is at Maxwell's tonight" I want to parse the piece of
it that talks about location, so Maxwell in this case.
If you say: "New asphalt will be done at 130E with 200S" then I want to
parse 103E and 200S.
So my question is:
- Without training, is there any algorithm that can identify "where" based
on the grammar structure?
Is there any paper related to that?
Wouldn't you typically use named entity recognition to detect the
location instead?
We have a location model for English. If that doesn't work well you
might need
to train one on your own data, since the provided one is trained on news
articles.
Jörn
