Re: Experiment: How good is quality of OpenNLP models for various languages.

[le...@interia.eu]

Hi Jeff

It's a good idea to have some pretrained models available to download.
I thought about setup automatic training for several languages. Maybe on gitlab? However I don't know exactly what they offer.
I will go back to this issue after holidays.

Regards
Leszek

Od: "Jeff Zemerick" <jz...@apache.org>
Do: users@opennlp.apache.org; 
Wysłane: 23:24 Środa 2022-07-27
Temat: Re: Experiment: How good is quality of OpenNLP models for various languages.

> Hi Leszek,
> 
> If you (or anyone else :) train any models that you would like to
share, we
> would be glad to see about making them available on the "Models
Download"
> page at https://opennlp.apache.org/models.html. Or we could create a
new
> page that links to models (such as your GitHub pages) developed by the
> community.
> 
> Thanks,
> Jeff
> 
> 
> On Sat, Jun 18, 2022 at 12:56 PM Alexandre Rademaker

> wrote:
> 
> > Good to know. Thank you.
> >
> > On Fri, 17 Jun 2022 at 07:03  wrote:
> >
> > > Hi Alexandre
> > > "Unfortunately" model training for portugese language went without
any
> > > problems
> > >
> > > In example:
> > > pt-pos-tagger.txt
> > > === EVALUATION INFO ===
> > > Evaluation-Score=0.9232609658839167
> > > Training-Sample-Size=8710
> > > Evaluation-Sample-Size=967
> > > Training-Algorithm=MAXENT
> > >
> > > pt-lemmatizer.txt
> > > === EVALUATION INFO ===
> > > Evaluation-Score=0.9815241470979176
> > > Training-Sample-Size=8710
> > > Evaluation-Sample-Size=967
> > > Training-Algorithm=MAXENT
> > >
> > --
> > Alexandre Rademaker
> > http://arademaker.github.com/
> > http://researcher.ibm.com/person/br-alexrad
> >
>