Skip to main content
. 2016 Oct 3;11(10):e0163480. doi: 10.1371/journal.pone.0163480

Table 4. F1-measure results for the global context kernel with combination of different feature sets: Negation scope and cue (N), Clause dependency (C), and neuTral candidate (T).

Category Global Context (%)
- +N +C +T +NC +CT +NT +NCT
DrugBank +Negation -Connector 56.6 54.9 58.6 66.2 57.8 67.2 59.8 62.1
+Connector 51.7 52.2 52.9 59.7 52.3 59.8 58.2 58.0
-Negation -Connector 64.7 64.8 64.8 71.8 64.8 71.9 71.9 72.9
+Connector 62.3 62.3 65.3 65.3 63.7 66.4 65.7 65.9
Total 61.7 61.3 62.9 68.6 62.4 69.0 67.5 68.3
MEDLINE +Negation -Connector 31.1 32.6 34.2 37.5 37.2 37.4 38.2 39.2
+Connector 28.3 33.5 33.5 35.2 38.4 35.8 39.4 38.2
-Negation -Connector 34.6 38.7 35.4 35.4 36.1 34.8 37.5 38.4
+Connector 34.1 38.2 36.7 37.2 35.4 36.4 39.5 40.6
Total 33.9 36.6% 34.2 36.0 36.8 36.7 38.4 38.8