ctakes-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Masanz, James J." <Masanz.Ja...@mayo.edu>
Subject RE: Difference between Refsem Class and TextSem class
Date Mon, 05 May 2014 16:23:53 GMT
1 The simplest is to run AgregateTemplateFiller engine, which is under ctakes-template-filler
To modifiy AggregatePlaintextprocessUMLS.xml to create the refsem elements, you’d basically
be re-creating AgregateTemplateFiller.xml

2. To get details of dosage and frequency, if you didn’t care about refsem, you would run
the DrugAggregatePlaintextUMLSProcessor, which is under ctakes-drug-ner. But the AgregateTemplateFiller
engine includes the drug ner engine, so once you switch to AgregateTemplateFiller, you should
see dosage and frequency

3. There’s some documentation about the components in

Suggestions welcome

-- James

From: Anirban Chakraborti [mailto:chakraborti.anirban@googlemail.com]
Sent: Monday, May 05, 2014 10:30 AM
To: user@ctakes.apache.org
Subject: Re: Difference between Refsem Class and TextSem class

thanks James,

1. I should be able to view refsem annotations via AggregatePlaintextprocessUMLS.xml if I
change the script to output refsem.element . Cant I? or do I have to call some another analysis

2. For a line like " Metronidazole 500 mg p.o. q.8h." , the textsem picks up the medication
and refers to its ontology concept( RXNOM code) but is not able to show dosage and frequency.
What should be done. Would refem be more useful.

3. Use case for each function , if available would have been great. are there any.


On Mon, May 5, 2014 at 8:44 PM, Masanz, James J. <Masanz.James@mayo.edu<mailto:Masanz.James@mayo.edu>>
org.apache.ctakes.typesystem.type.refsem.DiseaseDisorder  annotations are created by the TemplateFillerAnnotator
within ctakes-template-filler. Running AggregateTemplateFiller engine, which is under ctakes-template-filler,
will generate refsem annotations.

The concept behind having separate textsem and refsem annotations is if you have two textsem
annotations, they might refer to a single real-world object.  The goal is to (someday) have
a single refsem annotation per single real-world object, using coreference resolution to determine
which textsem annotations refer to the same thing, and having something to merge and resolve
conflicts between attributes of the coreferenced textsem annotations. That piece (merging
and dealing with conflicts of attributes) doesn’t exist yet.

So for now, the distinction between the refsem and textsem annotation is that refsem annotations
use the output of the relation extraction piece to set the bodyLocation and severity attributes.

-- James

From: Anirban Chakraborti [mailto:chakraborti.anirban@googlemail.com<mailto:chakraborti.anirban@googlemail.com>]
Sent: Monday, May 05, 2014 9:02 AM
To: user@ctakes.apache.org<mailto:user@ctakes.apache.org>
Subject: Difference between Refsem Class and TextSem class

what is the difference between refsem class and textsem class. How do I get an ouput like



View raw message