By the way, creating pipelines with code is "easier" if you use uimafit and avoid the old descriptor xml files.
As an aside, as of this weekend you can run the default fast pipeline and specify the dictionary configuration file with "-l pathToXml" on the command line. -----Original Message----- From: Martijn [mailto:mgkersl...@uvic.ca] Sent: Tuesday, March 14, 2017 1:36 PM To: dev@ctakes.apache.org Subject: Re: 2016AB UMLS (ctakessnorx) I have written a java class that creates the pipeline using the AggregatePlaintextFastUMLSProcessor.xml file. I changed the DictionaryDescriptorFile fileUrl parameter in the UmlsLookupAnnotator.xml to file:org/apache/ctakes/dictionary/lookup/fast/custom2.xml. So, as you can see, it uses the codingScheme custom2, but no code is returned. cough : UmlsConcept codingScheme: "custom2" code: <null> oid: "null#custom2" oui: <null> score: 0.0 disambiguated: false cui: "C0010200" tui: "T184" preferredText: "Coughing" On 14/03/2017, 10:24, "Finan, Sean" <sean.fi...@childrens.harvard.edu> wrote: How are you creating your pipeline? Do you run the command line script, run with an xml descriptor file in the CPE gui, or have you written a java class that creates the pipeline? DictionaryDescriptor points ctakes to your custom configuration file. Without it ctakes is probably defaulting to the 2011ab dictionary.