Detecting locations in text

["Kulkarni, Neeraj" <ne...@gmail.com>]

Hello,

I wanted to seek some help in identifying phrases in a sentence that
indicate a location. I tried using the OpenNLP en-location.bin model with
NER tagging, but it can detect well-defined locations, e.g. Atlanta, but
not location references.

For example, I was able to detect location (Atlanta) from sentence like
"Let's meet tomorrow in Atlanta" but couldn't find location (the atrium) in
the sentence "Let's meet tomorrow morning in the atrium"

Can someone guide me if NER is the right parsing technique for this, or if
I should start from a POS tagger and try to custom-identify phrases with
prepositions, proper nouns, etc. ? Or can OpenNLP be extended in some way
to create a parser that detects such references?

Thanks,
Neeraj