Skip to main content

Table 4 F1 score (precision, recall) of the baselines and the proposed method, named Combination

From: Learning adaptive representations for entity recognition in the biomedical domain

 

Single representation

Concatenation

Combination

Entity

SVM

NN

SVM

NN

MKL

NN

chebi

69.67

70.46

75.29

76.96

78.99

76.87

 

(88.84, 57.30)

(92.44,56.93)

(87.46, 66.08)

(83.91, 67.62)

(91.13, 69.70)

(90.1766.98)

cell

79.91

80.12

79.91

79.91

80.16

80.12

 

(88.41, 72.91)

(88.92, 72.91)

(88.41, 72.91)

(88.41, 72.91)

(89.01, 72.91)

(88.92, 72.91)

go_cc

66.50

65.31

67.92

65.56

68.92

65.59

 

(82.11, 55.87)

(81.92, 54.30)

(83.75, 57.12)

(84.31, 54.63)

(89.41, 56.06)

(82.81, 54.30)

go_bpmf

30.84

30.49

30.29

30.29

36.20

30.64

 

(68.59, 19.89)

(69.72, 19.51)

(70.96, 19.26)

(70.32, 19.30)

(78.73, 23.50)

(71.34, 19.51)

organism

92.36

92.97

92.85

93.32

94.99

93.27

 

(97.86, 87.45)

(99.24, 87.45)

(98.19, 88.07)

(99.35, 87.97)

(99.19, 91.13)

(99.19, 8.02)

protein

72.68

81.94

82.26

81.09

81.68

84.38

 

(77.63, 68.33)

(81.41, 82.48)

(84.40, 80.23)

(85.53, 77.08)

(85.44, 78.23)

(88.73, 80.43)

sequence

72.26

72.61

71.77

72.47

75.08

72.51

 

(89.58, 60.55)

(88.79, 61.42)

(88.67, 60.29)

(89.10, 61.07)

(93.11, 62.90)

(90.43, 60.52)

  1. Best results are highlighted in bold characters