Input Formats & Support Details

Simple Plain Text

Newlines for Sentences

Local sample configuration files (under config/):

  • plaintext_sentences.conf

Structured Plain Text (e.g., csv)

CSV with Start, End, and Negation Columns

  • csv_diagnoses.conf

If the configuration file includes a key/value pair for Opt Col, then we forcibly include the following three available values for this column:

  • affirmed
  • negated
  • possible

brat Annotation

The brat rapid annotation tool generates brat standoff format. Annotations are stored in a seconardy file (*.ann) while the original text is found in a plain text file (*.txt). This standoff format uses character offsets to locate spans: “All offsets all [sic] indexed from 0 and include the character at the start offset but exclude the character at the end offset.” See BioNLP Shared Task standoff format for a related format.

Limitations: The extraction engine currently only handles continous text-bound annotations for evaluation. Binary attributes can be extracted and included in the evaluation dictionary but are not scored themselves. Discontinous text-bound annotations, relations, events, multi-value attributes, normalizations, and notes are not supported.

Local sample configuration files (under config/):

  • brat_problems_allergies_standoff.conf

XML Formats


Local sample configuration files (under config/):

  • CAS_XMI.conf
  • i2b2_2016_track-1.conf
  • uima_sentences.conf
  • webanno_phi_xmi.conf
  • webanno_problems_allergies_xmi.conf
  • webanno_uima_xmi.conf


Extra sample configuration files (via the ETUDE engine configs repository):

  • i2b2/…
  • n2c2/n2c2_2018_track-1.conf