Table 6 Prompt strategy result Table.

From: An improved GPT2-based joint event extraction method with position expansion and knowledge augmentation

Model

Event Type Classification

Trigger Extraction

Argument Extraction

Mean F1

P

R

F1

P

R

F1

P

R

F1

QA_Format

93.70

95.01

94.35

88.86

90.03

89.44

79.93

80.96

80.44

88.07

Keywords

91.88

93.19

92.53

89.32

90.49

89.90

80.60

81.63

81.11

87.84

Natural

93.09

94.40

93.74

90.70

91.86

91.28

82.42

83.45

82.94

89.32

Ours

94.30

95.36

94.88

90.58

91.75

91.22

85.08

86.42

85.74

90.61

Back to article page