Skip to main content

Table 2 Model Performance Comparison

From: Multitask learning for biomedical named entity recognition with cross-sharing structure

  

Baseline Single-task Model (STM)

Fully-shared Multi-task Model (FS-MTM)

Shared-private Multi-task Model (SP-MTM)

Adversarial Multi-task Model (ADV-MTM)

Cross-sharing Multi-task Model (CS-MTM)

BC2GM

Precision

84.00

83.34

84.51

83.66

83.12

 

Recall

83.82

84.75

84.17

84.05

85.74

 

F1

83.91

84.04

84.34

83.85

84.41

Ex-PTM

Precision

70.83

72.56

70.45

76.60

74.73

 

Recall

64.12

70.46

70.03

67.43

69.56

 

F1

67.31

71.49

70.24

71.72

72.05

NCBI-disease

Precision

88.45

84.39

87.11

86.02

86.59

 

Recall

83.78

86.61

85.49

86.86

86.42

 

F1

86.05

85.49

86.29

86.44

86.50

Linnaeus

Precision

92.86

92.66

93.00

93.74

89.81

 

Recall

67.62

66.76

73.86

73.81

76.12

 

F1

78.25

77.60

82.33

82.59

82.40

  1. Bold: the best F1 score for the dataset