You don't really have any control over how the machine-learning based negation detection algorithm works. Apparently, that algorithm was trained on a corpus where '-' represented a negative.
If you use the negex-based negation detection algorithm, you have much more control over negation detection (you can define your own triggers). Try ctakes-ytex-uima\desc\analysis_engine\NegexAnnotator.xml instead of the ctakes assertion analysis engine. HTH VJ On Tue, Oct 7, 2014 at 7:40 PM, Kim Ebert <[email protected]> wrote: > Hi Natalia, > > Here are a few variations on your sentence with the results. > > The following results in a polarity of 1 for all of the > DiseaseDisorderMentions. "Unspecified pervasive developmental disorder, > active state" > > <org.apache.ctakes.typesystem.type.textsem.DiseaseDisorderMention > _indexed="1" _id="180" _ref_sofa="1" begin="0" end="44" id="0" > _ref_ontologyConceptArr="173" typeID="2" segmentID="SIMPLE_SEGMENT" > discoveryTechnique="1" confidence="1.0" polarity="1" uncertainty="0" > conditional="false" generic="false" subject="patient" historyOf="0"/> > <org.apache.ctakes.typesystem.type.textsem.DiseaseDisorderMention > _indexed="1" _id="233" _ref_sofa="1" begin="22" end="44" id="2" > _ref_ontologyConceptArr="229" typeID="2" segmentID="SIMPLE_SEGMENT" > discoveryTechnique="1" confidence="1.0" polarity="1" uncertainty="1" > conditional="false" generic="false" subject="patient" historyOf="0"/> > <org.apache.ctakes.typesystem.type.textsem.DiseaseDisorderMention > _indexed="1" _id="276" _ref_sofa="1" begin="36" end="44" id="3" > _ref_ontologyConceptArr="273" typeID="2" segmentID="SIMPLE_SEGMENT" > discoveryTechnique="1" confidence="1.0" polarity="1" uncertainty="1" > conditional="false" generic="false" subject="patient" historyOf="0"/> > <org.apache.ctakes.typesystem.type.textsem.DiseaseDisorderMention > _indexed="1" _id="359" _ref_sofa="1" begin="12" end="44" id="1" > _ref_ontologyConceptArr="352" typeID="2" segmentID="SIMPLE_SEGMENT" > discoveryTechnique="1" confidence="1.0" polarity="1" uncertainty="0" > conditional="false" generic="false" subject="patient" historyOf="0"/> > > "Unspecified pervasive developmental disorder,* current or *active state" > results with one negative polarity. > > <org.apache.ctakes.typesystem.type.textsem.DiseaseDisorderMention > _indexed="1" _id="204" _ref_sofa="1" begin="0" end="44" id="0" > _ref_ontologyConceptArr="197" typeID="2" segmentID="SIMPLE_SEGMENT" > discoveryTechnique="1" confidence="1.0" polarity="1" uncertainty="0" > conditional="false" generic="false" subject="patient" historyOf="0"/> > <org.apache.ctakes.typesystem.type.textsem.DiseaseDisorderMention > _indexed="1" _id="257" _ref_sofa="1" begin="22" end="44" id="2" > _ref_ontologyConceptArr="253" typeID="2" segmentID="SIMPLE_SEGMENT" > discoveryTechnique="1" confidence="1.0" polarity="1" uncertainty="1" > conditional="false" generic="false" subject="patient" historyOf="0"/> > <org.apache.ctakes.typesystem.type.textsem.DiseaseDisorderMention > _indexed="1" _id="300" _ref_sofa="1" begin="36" end="44" id="3" > _ref_ontologyConceptArr="297" typeID="2" segmentID="SIMPLE_SEGMENT" > discoveryTechnique="1" confidence="1.0" polarity="1" uncertainty="1" > conditional="false" generic="false" subject="patient" historyOf="0"/> > <org.apache.ctakes.typesystem.type.textsem.DiseaseDisorderMention > _indexed="1" _id="383" _ref_sofa="1" begin="12" end="44" id="1" > _ref_ontologyConceptArr="376" typeID="2" segmentID="SIMPLE_SEGMENT" > discoveryTechnique="1" confidence="1.0" polarity="*-1*" uncertainty="0" > conditional="false" generic="false" subject="patient" historyOf="0"/> > > > "* - *Unspecified pervasive developmental disorder, current or active > state* - *" results in three negative polarities. > > <org.apache.ctakes.typesystem.type.textsem.DiseaseDisorderMention > _indexed="1" _id="180" _ref_sofa="1" begin="39" end="47" id="3" > _ref_ontologyConceptArr="177" typeID="2" segmentID="SIMPLE_SEGMENT" > discoveryTechnique="1" confidence="1.0" polarity="1" uncertainty="0" > conditional="true" generic="false" subject="patient" historyOf="0"/> > <org.apache.ctakes.typesystem.type.textsem.DiseaseDisorderMention > _indexed="1" _id="263" _ref_sofa="1" begin="3" end="47" id="0" > _ref_ontologyConceptArr="256" typeID="2" segmentID="SIMPLE_SEGMENT" > discoveryTechnique="1" confidence="1.0" polarity="*-1*" uncertainty="0" > conditional="false" generic="false" subject="patient" historyOf="0"/> > <org.apache.ctakes.typesystem.type.textsem.DiseaseDisorderMention > _indexed="1" _id="346" _ref_sofa="1" begin="15" end="47" id="1" > _ref_ontologyConceptArr="339" typeID="2" segmentID="SIMPLE_SEGMENT" > discoveryTechnique="1" confidence="1.0" polarity="*-1*" uncertainty="0" > conditional="false" generic="false" subject="patient" historyOf="0"/> > <org.apache.ctakes.typesystem.type.textsem.DiseaseDisorderMention > _indexed="1" _id="399" _ref_sofa="1" begin="25" end="47" id="2" > _ref_ontologyConceptArr="395" typeID="2" segmentID="SIMPLE_SEGMENT" > discoveryTechnique="1" confidence="1.0" polarity="*-1*" uncertainty="0" > conditional="false" generic="false" subject="patient" historyOf="0"/> > > "* - *Unspecified pervasive developmental disorder, current or active > state" results are the same. > <org.apache.ctakes.typesystem.type.textsem.DiseaseDisorderMention > _indexed="1" _id="172" _ref_sofa="1" begin="39" end="47" id="3" > _ref_ontologyConceptArr="169" typeID="2" segmentID="SIMPLE_SEGMENT" > discoveryTechnique="1" confidence="1.0" polarity="1" uncertainty="1" > conditional="false" generic="false" subject="patient" historyOf="0"/> > <org.apache.ctakes.typesystem.type.textsem.DiseaseDisorderMention > _indexed="1" _id="255" _ref_sofa="1" begin="3" end="47" id="0" > _ref_ontologyConceptArr="248" typeID="2" segmentID="SIMPLE_SEGMENT" > discoveryTechnique="1" confidence="1.0" polarity="*-1*" uncertainty="0" > conditional="false" generic="false" subject="patient" historyOf="0"/> > <org.apache.ctakes.typesystem.type.textsem.DiseaseDisorderMention > _indexed="1" _id="338" _ref_sofa="1" begin="15" end="47" id="1" > _ref_ontologyConceptArr="331" typeID="2" segmentID="SIMPLE_SEGMENT" > discoveryTechnique="1" confidence="1.0" polarity="*-1*" uncertainty="0" > conditional="false" generic="false" subject="patient" historyOf="0"/> > <org.apache.ctakes.typesystem.type.textsem.DiseaseDisorderMention > _indexed="1" _id="391" _ref_sofa="1" begin="25" end="47" id="2" > _ref_ontologyConceptArr="387" typeID="2" segmentID="SIMPLE_SEGMENT" > discoveryTechnique="1" confidence="1.0" polarity="*-1*" uncertainty="0" > conditional="false" generic="false" subject="patient" historyOf="0"/> > > > Oddly, "Unspecified pervasive developmental disorder, current or active > state -" results in one negative polarity. > > <org.apache.ctakes.typesystem.type.textsem.DiseaseDisorderMention > _indexed="1" _id="212" _ref_sofa="1" begin="0" end="44" id="0" > _ref_ontologyConceptArr="205" typeID="2" segmentID="SIMPLE_SEGMENT" > discoveryTechnique="1" confidence="1.0" polarity="*-1*" uncertainty="0" > conditional="false" generic="false" subject="patient" historyOf="0"/> > <org.apache.ctakes.typesystem.type.textsem.DiseaseDisorderMention > _indexed="1" _id="265" _ref_sofa="1" begin="22" end="44" id="2" > _ref_ontologyConceptArr="261" typeID="2" segmentID="SIMPLE_SEGMENT" > discoveryTechnique="1" confidence="1.0" polarity="1" uncertainty="0" > conditional="true" generic="false" subject="patient" historyOf="0"/> > <org.apache.ctakes.typesystem.type.textsem.DiseaseDisorderMention > _indexed="1" _id="308" _ref_sofa="1" begin="36" end="44" id="3" > _ref_ontologyConceptArr="305" typeID="2" segmentID="SIMPLE_SEGMENT" > discoveryTechnique="1" confidence="1.0" polarity="1" uncertainty="0" > conditional="true" generic="false" subject="patient" historyOf="0"/> > <org.apache.ctakes.typesystem.type.textsem.DiseaseDisorderMention > _indexed="1" _id="391" _ref_sofa="1" begin="12" end="44" id="1" > _ref_ontologyConceptArr="384" typeID="2" segmentID="SIMPLE_SEGMENT" > discoveryTechnique="1" confidence="1.0" polarity="1" uncertainty="0" > conditional="true" generic="false" subject="patient" historyOf="0"/> > > I hope this helps. > > > Kim Ebert1.801.669.7342 > Perfect Search Corphttp://www.perfectsearchcorp.com/ > > On 09/30/2014 09:14 AM, Natalia Connolly wrote: > > Dear cTAKES Experts, > > I have a piece of free text that includes a diagnosis in a stand-alone > sentence, like this: > > " - Unspecified pervasive developmental disorder, current or active > state - " > > For some reason cTAKES seems to think the polarity of this > statement is negative: > > <org.apache.ctakes.assertion.medfacts.types.Concept _indexed="1" > _id="18158" _ref_sofa="3" begin="936" end="980" conceptType="PROBLEM" > conceptText="Unspecified pervasive developmental disorder" externalId="0" > originalEntityExternalId="8563"/> > > <org.apache.ctakes.typesystem.type.textsem.DiseaseDisorderMention > _indexed="1" _id="8563" _ref_sofa="3" begin="936" end="980" id="40" > _ref_ontologyConceptArr="8556" typeID="2" segmentID="SIMPLE_SEGMENT" > discoveryTechnique="1" confidence="1.0" *polarity="-1"* uncertainty="0" > conditional="false" generic="false" subject="patient" historyOf="0"/> > > Why is that?? Can it be the hyphens? > > Thanks for any insight, > > Natalia Connolly > > >
