The coref code was moved to the sandbox because the lack of maintenance
blocked the further development of the machine learning code.
JWNL was distributed with OpenNLP until 1.5.3, I suggest you just include
the library in your project directly. The coref code contained a small
util class
to lookup the lemmas.
Anyway we started a discussion a while ago to add a lemmatizer to OpenNLP,
if you are interested in this you are welcome to help us out there. We
would like
to define a common interface and then provide different implementations,
e.g.
one could be based on JWNL.
HTH,
Jörn
On 09/18/2013 10:58 AM, Ioan Barbulescu wrote:
Hi guys
I tried the 1.6.0 snapshot yesterday and I noticed that the coref codebase
is completely gone from 1.6 (it is still present in 1.5.3).
I know that code is rather old and I don't need all of it.
But I do need the JWNL part, for a very specific reason:
It is the only way to obtain the lemmas for the words (for example retrieve
"big" from "bigger" or "eat" from "eats"). Or, at least, it is the only way
I found so far.
So, my questions are the following:
- are there any other ways (more actual / recommended) ways to retrieve the
lemmas from the input words? (which I somehow didn't find?)
- if not, would be possible to keep at least the JWNL code in 1.6?
In the defense of this proposal, I confirm that the JWNL code (in
particular) works just well with the rest of the 1.6 codebase. I stripped
all of the coref .class code from the 1.5.3 jar and I left only the
JWNL-related bytecode. I successfully run this code, together with the
1.6.0-snapshot jar.
Many thanks.
BR,
Ioan
