uima-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Richard Eckart de Castilho <...@apache.org>
Subject Re: UIMAFit vs. LEO
Date Wed, 03 Jun 2015 16:39:24 GMT
Hi all,

I think that LEO and uimaFIT are largely orthogonal.

While both try to make the use of UIMA easier, LEO focusses on
UIMA-AS while uimaFIT focusses on the core UIMA API.

There appear to be some overlaps between LEO and uimaFIT,
e.g. when it comes to generating descriptors. LEO appears
to be using a more heavy-weight approach by providing its
own descriptor wrapper classes where the uimaFIT factories
provide follow what I would see as a more lightweight approach.

I think it should be possible to mix uimaFIT and LEO in such
a sense that descriptors created with uimaFIT can be consumed
by LEO and uimaFIT-aware components should run nicely on LEO.

Cheers,

-- Richard

On 21.05.2015, at 17:19, Thomas Ginter <thomas.ginter@utah.edu> wrote:

> Hi Peter,
> 
> That is an excellent question.  Leo was written with two goals in mind.  You have touched
on one of them.  
> 
> 1.  Utilize UIMA-AS to lend scalability to our NLP processing.
> 2.  Facilitate the reuse of existing and third party code.
> 
> Obviously it was built as a means of programmatically creating and managing pipelines
as UIMA-AS services, as well as managing the client side of UIMA-AS, to facilitate the first
goal.  Less well known is that because of the second goal Leo allows you to either utilize
the framework to create the description for an analysis engine or aggregate engine (pipeline)
but it also allows you to import descriptors or description objects generated from other sources.
 In this way we can reuse modules, or even whole pipelines, like YTEX and cTAKES, without
needing to have access to the source code to insert the appropriate metadata.  If you wanted
to use UIMAFit to generate the description object for an analysis engine and then insert that
analysis engine into your Leo pipeline it is also relatively easy to do so. 
> 
> Thanks,
> 
> Thomas Ginter
> 801-448-7676
> thomas.ginter@utah.edu
> 
> 
> 
> 
>> On May 19, 2015, at 18:49, Petr Baudis <pasky@ucw.cz> wrote:
>> 
>> Hi!
>> 
>> On Thu, May 14, 2015 at 05:44:12PM +0000, Thomas Ginter wrote:
>>> There is also Leo which allows you to programmatically create pipelines, launch
them as UIMA-AS services, and manage types systems and clients without having to touch any
descriptor files.  You can find documentation at the site below:
>>> 
>>> http://department-of-veterans-affairs.github.io/Leo/userguide.html
>> 
>> I'm wondering how does UIMAFit and LEO fit together.  My impression
>> right now is:
>> 
>> * They both have the same goal.
>> 
>> * Mixing them in the same pipeline might get messy(?)
>> 
>> * LEO advantage is that it seamlessly works with UIMA-AS (in fact it's
>>   built around UIMA-AS).
>> 
>> * UIMAFit advantage is (if nothing else) vastly wider ecosystem.
>> 
>> Did I get this about right?
>> 
>> Thanks,
>> 
>> 				Petr Baudis
> 


Mime
View raw message