Skip to main content
. Author manuscript; available in PMC: 2016 Aug 17.
Published in final edited form as: J Biomed Inform. 2015 Jun 26;58(Suppl):S111–S119. doi: 10.1016/j.jbi.2015.06.010

Table 4. Features used in ML classifiers.

All classifiers

F1 Indexed uncased previous words
F2 Indexed uncased next words
F3 Generic words within 5 tokens
F4 Has family term within 5 tokens
F5 Negation word in previous 10 tokens
F6 Modality word in previous 10 tokens
F7 ConText negation value
F8 ConText history value
F9 ConText hypothetical value
F10 ConText experiencer value
F11 Section name
Measurement classifiers only Time classifiers only


F12 Words between base and value F18 Annotation type
F13 Word shapes between base and value F19 Medication type
F14 Value shape
F15 Base and value on same line
F16 Number of tokens between base and value
F17 Target word in previous 5 tokens