The (separately downoadable) dictionaries for anatomical sites, procedures, signs/symptoms, and disorders/diseases dictionaries are built from SNOMED-CT, ICD-9 and others, and cTAKES can assign UMLS Metathesaurus CUIs and SNOMED-CT codes for from those dictionaries.
Medications can be annotated with RXNorm codes and CUIs. Those annotations are all described in A common type system for clinical natural language processing <https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3575354/> You can build your own dictionaries from other vocabularies within the UMLS Metathesaurus using the Dictionary Creator GUI <https://cwiki.apache.org/confluence/display/CTAKES/Dictionary+Creator+GUI>. Semantic role labelling follows the conventions of PropBank <https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3575354/#B15> Temporal annotations are based on TimeML: http://www.aclweb.org/anthology/S13-2002 http://clear.colorado.edu/compsem/documents/THYME_guidelines.pdf hope that helps -- James Masanz On Tue, Oct 24, 2017 at 1:14 PM, Marcello Bax UFMG <[email protected]> wrote: > Hi, > Could anyone clarify me if cTAKES-generated annotations follow some > standard format or ontology, like Web Annotation Ontology (W3C), BioC, and > others? > It would facilitate interoperability issues. > > Thanks, > Marcello Bax > Information Theory and Management Department - Information Science School > - UFMG > bax.eci.ufmg.br >
