uima-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Peter Klügl <peter.klu...@averbis.com>
Subject Re: How to get TextRuler to work
Date Thu, 14 Feb 2019 10:37:30 GMT
Hi,

Am 14.02.2019 um 09:44 schrieb Mandy Neumann:
> Hi,
>
> yes I already read elsewhere that TextRuler is not actively
> maintained. Which is a pity because it has the potential to be a
> really cool tool! We really need something like that in our project.
>

I could talk for hours about black-box vs white-box machine learning vs
manual rule engineering also concerning the maintanance of the annotator
and developement cycles...

Afterall, it's a question of priorities where the available time will be
spent and even if TextRuler is cool, it was always at the bottom of my
priority list for several years.


> Thanks for pointing me to the error log, I eventually found the
> problem in a NullPointerException - Ruta was not able to find the
> script Base.ruta I had imported in Features.ruta (I wanted to mirror
> the example project as closely as possible). Although both reside in
> the same package, it was not clear to me that I still have to use the
> fully qualified name to import.
>

Glad to hear that you found the problem.


> By the way, in case anybody wants to pick up maintaining TextRuler
> again, I would suggest to improve a bit on error handling here.


You can also help :-)

Just by creating a bug report in our Jira, you will put it on my todo
list. Also if I do not find the time to fix it, maybe others will.

https://issues.apache.org/jira/browse/UIMA-5987?jql=component%20%3D%20ruta


Best,


Peter



>
> Cheers,
>
> Mandy
>
>
> Am 12.02.19 um 16:48 schrieb Peter Klügl:
>> Hi,
>>
>>
>> I haven't used TextRuler for years and therefore I cannot tell right
>> away what the issue could be. I also have to mention that the view and
>> its algorithms are not actively maintained.
>>
>>
>> Is there a message in the Error Log of Eclipse?
>>
>>
>> Can you provide a minimal exmaple for reproducing the problem?
>>
>>
>> Best,
>>
>>
>> Peter
>>
>>
>> Am 11.02.2019 um 17:22 schrieb Mandy Neumann:
>>> Hi,
>>>
>>> after fixing my initial problems with the CAS views, I now want to
>>> proceed to my main task. I want to apply TextRuler to learn a set of
>>> rules instead of writing them manually.
>>>
>>> Unfortunately, I can't get TextRuler to work in my project. The
>>> example project works, but when I start the process on my project, the
>>> view displays "Preprocessing... Loading XMI file (input): <file>" and
>>> does nothing more. I also cannot stop, pressing the stop button has no
>>> effect, I need to restart the workbench to be able to change TextRuler
>>> settings.
>>>
>>> I'm not sure if my workflow is even right. The first step is to
>>> convert html input into XMIs with the html tags as annotations. In
>>> order to create the training data, I figured I need to include the
>>> target type system already at this step. I then used the
>>> "Annotate/Quick Annotate" options in the CAS viewer to create the gold
>>> standard annotations.
>>>
>>> I also created two Ruta scripts to be used by TextRuler. Base.ruta
>>> basically just declares the target type system. Features.ruta creates
>>> some basic annotations that should be used by TextRuler to infer rules.
>>>
>>> Anybody an idea why my workflow won't work?
>>>
>>> Cheers,
>>>
>>> Mandy
>>>
>>>
-- 
Dr. Peter Klügl
R&D Text Mining/Machine Learning

Averbis GmbH
Tennenbacher Str. 11
79106 Freiburg
Germany

Fon: +49 761 708 394 0
Fax: +49 761 708 394 10
Email: peter.kluegl@averbis.com
Web: https://averbis.com

Headquarters: Freiburg im Breisgau
Register Court: Amtsgericht Freiburg im Breisgau, HRB 701080
Managing Directors: Dr. med. Philipp Daumke, Dr. Kornél Markó


Mime
View raw message