Mailing-List: contact dev-help@opennlp.apache.org; run by ezmlm
Precedence: bulk
Reply-To: dev@opennlp.apache.org
MIME-Version: 1.0
In-Reply-To: <74845922-9BB6-4377-9CC6-1047EE8F0E9D@nih.gov>
References: <1486402832.2628.1.camel@gmail.com> <74845922-9BB6-4377-9CC6-1047EE8F0E9D@nih.gov>
From: Damiano Porta <damianoporta@gmail.com>
Date: Tue, 7 Feb 2017 15:51:47 +0100
Message-ID: <CALxjdk0xm0Gb6=5yVgjS6xSmt7CF6bVVMUwnrPxRwomCH6_qXQ@mail.gmail.com>
Subject: Re: Name Finder trainer default settings
To: dev@opennlp.apache.org
Content-Type: multipart/alternative; boundary=001a113d6e7232bb390547f1e392
archived-at: Tue, 07 Feb 2017 14:51:52 -0000

--001a113d6e7232bb390547f1e392
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

I have good results with perceptron, but +1 for CRF

2017-02-07 15:42 GMT+01:00 Russ, Daniel (NIH/CIT) [E] <druss@mail.nih.gov>:

> Hi J=C3=B6rn,
>
>
>
>    I think the best entity recognition systems use CRF=E2=80=99s.  At som=
e point
> we might want to consider adding them.  As you know, ME classifiers suffe=
r
> from label bias problem (see Lafferty et. al<http://repository.upenn.
> edu/cgi/viewcontent.cgi?article=3D1162&context=3Dcis_papers>.) CRF=E2=80=
=99s deal
> with that issue.  I believe that perceptrons suffer from the same problem=
.
> If you think the results are better, I have no problem.  I think that our
> long-term goal should be to add a CRF, and make it the default for the
> NameFinder.
>
>
>
> Daniel
>
>
>
>
>
> On 2/6/17, 12:40 PM, "Joern Kottmann" <kottmann@gmail.com> wrote:
>
>
>
>     Hello all,
>
>
>
>     I would like to propose to switch the default training algorithm from
>
>     maxent gis to perceptron for the Name Finder. In all the data sets I
>
>     tried perceptron performs better than maxent gis and I believe that
>
>     would be a much more sensible default.
>
>
>
>     A user can always override the default by providing the algorithms
>
>     parameter for training.
>
>
>
>     What do you think?
>
>
>
>     J=C3=B6rn
>
>
>
>
>

--001a113d6e7232bb390547f1e392--