Skip to main content

Table 6 Vocabulary build steps required for each system

From: NOBLE – Flexible concept recognition for large-scale biomedical natural language processing

Concept Recognition Systems Dictionary Data Structure Used by Coder
MMTxa Used MetamorphoSys to convert RRF to ORF and used bundled data file builder to create terminology for each corpus; this process required significant user interaction and took many hours
MGrep Sent RRF files for both corpora to the MGrep authors and received from them a tab delimited text file that could be used with the MGrep system enriched with LVG; there is limited publicly available information about the vocabulary format required by MGrep
Concept Mapper Wrote custom Java code to convert RRF files to an XML file formatted in the Concept Mapper valid syntax
cTAKES Dictionary Lookup Annotator Wrote custom Java code to convert RRF files to seed a Lucene Index
cTAKES Fast Dictionary Lookup Annotator Wrote custom Java code to convert RRF into Bar Separated Values (BSV) file that FDLA imports
NOBLE Codera Directly imported RRF files
  1. aSystems that have vocabulary import and selection tooling