On 1/4/11 11:08 AM, Rao, Vaijanath wrote:
Hi Sriram,
There are many ways you can accomplish this via opennlp. One way is to treat
them as Named Entities and then create a named entity training corpus first for
those entities.
Once you have created the named entity model, you can then use that to identify
these terms. You might have to modify the feature generator to suite your
requirements.
I worked on a similar task a while back and extended the feature
generators of the name finder a little
to be able to detect these entities better.
We are still missing a component to disambiguate/identify names, which I
think you also need.
Jörn
