Skip to main content

Table 1 Performance of STL-DS and MTL-XC on all tasks

From: Hierarchical shared transfer learning for biomedical named entity recognition

Dataset STL MTL-XC
P. % R. % F1 P. % R. % F1
Chemical
BC4CHEMD 93.00 92.40 91.70 92.02 92.49 91.25
BC5CDR 92.76 93.96 93.36 93.43 92.94 93.19
BioNLP11ID 55.56 72.58 62.94 61.44 75.81 67.87
BioNLP13CG 83.20 85.14 84.16 82.35 80.88 81.61
BioNLP13PC 88.57 90.33 89.44 76.65 83.80 80.07
CRAFT 84.07 81.05 82.54 75.34 72.38 73.83
Disease
BC5CDR 84.83 88.11 86.44 86.40 87.34 86.87
NCBI-disease 87.27 89.27 88.26 87.80 89.73 88.75
Gene and protein
BC2GM 81.91 82.53 82.22 82.67 82.61 82.64
BioNLP09 88.20 86.82 87.50 87.23 91.95 89.53
BioNLP11EPI 84.23 87.96 85.81 85.32 87.63 86.46
BioNLP11ID 89.22 89.65 89.43 89.6 88.47 89.03
BioNLP13CG 88.45 92.42 90.39 93.56 91.63 92.58
BioNLP13GE 73.57 83.51 78.22 77.62 90.55 83.59
BioNLP13PC 89.56 94.26 91.85 90.66 87.93 89.27
CRAFT 80.48 75.44 77.88 78.56 84.83 81.57
Ex-PTM 74.79 80.46 77.52 81.83 86.83 84.25
JNLPBA 71.98 80.04 75.80 72.58 85.04 78.32
Species
BioNLP11ID 85.41 82.03 83.68 91.22 70.24 79.37
BioNLP13CG 88.34 89.19 88.76 88.39 86.68 87.52
CRAFT 96.45 97.73 97.08 93.76 93.51 93.63
LINNAEUS 91.70 85.62 88.56 88.43 82.14 85.17
  1. Better scores of each metric are in bold