Skip to main content

Table 5 The effect analysis of each feature/strategy

From: Disease named entity recognition from biomedical literature using a novel convolutional neural network

Corpus

Feature/strategy

P

R

F-score

Δ

NCBI

None

85.08

85.26

85.17

–

Character-level

84.50

84.41

84.46

0.71

Lexicon

83.53

83.35

83.44

1.73

MLS

83.97

84.41

84.19

0.98

Post-processing

84.84

84.84

84.84

0.33

CDR

None

88.20

87.46

87.83

–

Character-level

87.25

87.35

87.30

0.53

Lexicon

85.48

84.90

85.19

2.64

MLS

87.18

86.17

86.67

1.16

Post-processing

86.67

86.69

86.68

1.15

  1. Notes. Δdenotes the corresponding F-score decrease when a strategy or a feature is removed