*not* speaking on behalf of the team but I wrote my own little tool to
do the annotation whenever there is a dictionary or dictionaries
available. It is open-source, it exposes a Java and a clojure API or you
can even run it from the command line. You can find it here:
https://github.com/jimpil/annotator-clj
proper annotation in a specialised domain however is usually done by
domain experts. Don't expect to get gold-data using annotator-clj or in
fact any dictionary based annotator...the good with annotator-clj is
that it can produce annotations compatible with openNLP, stanfordNLP or
NLTK which is great sometimes. It is also very fast as it leverages the
fork-join framework available in java7.
hope that helps :)
Jim
ps: now that you've made me think about it again, instead of only
allowing raw dictionaries I could add support for the user to provide
his own specialised regex Pattern objects...hmmm...I'll investigate over
the weekend.
On 06/08/13 22:23, Girivaraprasad Nambari wrote:
Hi Team,
I ran pre-trained English NER models on text I have and results are not as
I expected. So, I thought to train NER on the "trainining" text to generate
new models, but I am stuck on *Annotation tools*.
Could someone help me on which tools are being used by Open-NLP team for
text annotation? As well as any references around text annotation topics
would be really appreciated.
Thanks for your time and help.
Thank you,
Giri
On Tue, Aug 6, 2013 at 5:12 PM, Girivaraprasad Nambari <
[email protected]> wrote:
Hi Team,
I ran pre-trained English NER models on text I have and results are not as
I expected. So, I thought to train NER on the "trainining" text to generate
new models, but I am stuck on *Annotation tools*.
Could someone help me on which tools are being used by Open-NLP team for
text annotation? As well as any references around text annotation topics
would be really appreciated.
Thanks for your time and help.
Thank you,
Giri
