You are viewing a plain text version of this content. The canonical link for it is here.
Posted to user@ctakes.apache.org by Jessica Glover <gl...@gmail.com> on 2015/12/31 15:52:57 UTC

Achieving YTEX negation in CuisOnlyPlaintextUMLSProcessor

I have noticed that using the AggregatePlaintextUMLSProcessor in
ctakes-ytex-uima results in improved negation detection over the one in
ctakes-clinical-pipeline on my test documents.

I would like to get the same improvement in CuisOnlyPlaintextUMLSProcessor.
I have tried replacing the DictionaryLookupAnnotator, which seemed to be
the determining factor in AggregatePlaintextUMLSProcessor, but when I run
it, I find no IdentifiedAnnotations. Does the YTEX
DictionaryLookupAnnotator depend on an analysis engine I've overlooked? Any
other suggestions?

Thank you,
Jessica G.

Re: Achieving YTEX negation in CuisOnlyPlaintextUMLSProcessor

Posted by vijay garla <vn...@gmail.com>.
CTAKES_HOME\desc\ctakes-ytex-uima\desc\analysis_engine\DictionaryLookupAnnotator.xml
references
CTAKES_HOME\resources\org\apache\ctakes\ytex\dictionary\lookup\LookupDesc_SNOMED.xml,
which limits all dictionary lookups
to org.apache.ctakes.typesystem.type.textspan.LookupWindowAnnotation spans,
which are created by the LookupWindowAnnotator.

You can either add the LookupWindowAnnotator (not sure about its
dependencies), or you can change LookupDesc_SNOMED.xml to use a different
annotation (e.g. Noun Phrase or Sentence).

HTH,

VJ

On Thu, Dec 31, 2015 at 3:52 PM, Jessica Glover <gl...@gmail.com>
wrote:

> I have noticed that using the AggregatePlaintextUMLSProcessor in
> ctakes-ytex-uima results in improved negation detection over the one in
> ctakes-clinical-pipeline on my test documents.
>
> I would like to get the same improvement in
> CuisOnlyPlaintextUMLSProcessor. I have tried replacing the
> DictionaryLookupAnnotator, which seemed to be the determining factor in
> AggregatePlaintextUMLSProcessor, but when I run it, I find no
> IdentifiedAnnotations. Does the YTEX DictionaryLookupAnnotator depend on an
> analysis engine I've overlooked? Any other suggestions?
>
> Thank you,
> Jessica G.
>