Skip to main content

Table 3 Performance comparison between the state of the art [4, 28] and this work in terms of precision (P), recall (R) and F-measure (F)

From: Syntax-based transfer learning for the task of biomedical relation extraction

Test corpus Work (train corpus) P R F
SNPPhena [4] (SNPPhenA) 56.6 59.8 58.2
  This work (SNPPhenA + SemEval 2013 DDI) 64.5 75.2 69.4
EU-ADR [28] (EU-ADR drug-disease) 70.2 93.2 79.3
drug-disease This work (EU-ADR drug-disease + SemEval 2013 DDI) 74.8 90.6 82.0
EU-ADR [28] (EU-ADR drug-target) 74.2 97.4 83.3
drug-target This work (EU-ADR drug-target + SemEval 2013 DDI) 73.5 95.6 83.1
EU-ADR [28] (EU-ADR target-disease) 75.1 97.7 84.6
target-disease This work (EU-ADR target-disease + SemEval 2013 DDI) 78.7 91.4 84.6
  1. Results reported for this work are ensembles of the 5 best models obtained. Bold numbers correspond to the best performing models