ctakes-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Bonnie MacKellar <macke...@stjohns.edu>
Subject RE: RUTA with cTakes?
Date Mon, 12 Oct 2015 12:19:56 GMT

Thanks for the heads up on CALL vs EXEC. I was looking at the RUTA docs, where it says "Additional
Analysis Engines can be imported with the keyword "ENGINE" followed by the name of the descriptor.
These imported Analysis Engines can be executed with the actions "CALL" or "EXEC"." - so I
thought they were interchangeable. I am very new to RUTA.

I will definitely try this through Java. I am having trouble correctly setting up cTakes projects
in Eclipse, but I will eventually figure it out. It looks like there are not a lot of people
trying to integrate these.

Bonnie MacKellar
Division of Computer Science, Math and Science
St John's University
Queens, NY

From: Peter Kl├╝gl [mailto:peter.kluegl@averbis.com]
Sent: Friday, October 09, 2015 8:51 AM
To: user@ctakes.apache.org
Subject: Re: RUTA with cTakes?


I have not done this yet, but I will soon be able to answer this question.

Right now, I can only give some pointers:

- You can use a ruta script for specifying a pipeline, but I would not recommend it in this
use case. There can maybe occur problems caused my conventions in cTAKES components that are
not considered in ruta (e.g., special resource loading). I do not have an overview yet.
- Analysis engine should rather be called with EXEC. The CALL action works better for imported
ruta scripts. The CALL action tries to consider the current filtering settings so that a java
analysis engine can be faced with a document text without whitespaces. The CALL action does
not support reindexing (I think) and no different CAS views, but the EXEC action does.
- If the pipeline and the analysis engine of cTAKES is hidden behind some interfaces, yes,
then it is better to call the ruta analysis engine on plain java level (right now, you still
need to consider ruta's type priorities).
- Without known cTAKES yet, I would recommend to include the ruta analysis engine (description)
in the pipeline/AAE together with the other components. Here, you probably face the fewest
problems, if any at all.

There are different approaches for setting up the ruta descriptors. Again, I probably need
a closer look at how cTAKES normally does it before I can give a good advice here. In case
you need support for ruta descriptors in general, you can always ask questions, e.g, on the
uima user list.



Am 09.10.2015 um 00:09 schrieb Bonnie MacKellar:

Has anyone successfully integrated the RUTA language with cTakes? I want to use cTakes to
do the UMLS annotation, and then feed the output to a RUTA script with rules for identifying
particular patterns.  I am really not sure how to approach this. Should I load the Annotator
into my RUTA script as in this example
TYPESYSTEM org.apache.ctakes.drugner.types.TypeSystem;
ENGINE DrugMentionAnnotator;

Document { -> CALL ( DrugMentionAnnotator)};

Or should I set things up in Java and call a Ruta script programmatically?  I am very unclear
as to the mechanics of tying these systems together, especially setting up the descriptors.


Bonnie MacKellar
Division of Computer Science, Math and Science
St John's University
Queens, NY

View raw message