Table 5 Performance comparison of our method using different BERTs and Glove embeddings.

From: Exploiting question-answer framework with multi-GRU to detect adverse drug reaction on social media

Methods

PSB2016

SMM4H2018

P(%)

R(%)

F(%)

P(%)

R(%)

F(%)

Our model

71.52

75.15

73.29

78.21

84.65

81.30

BERT32

73.08

68.53

70.73

77.0

78.90

77.94

BioBERT43

68.42

73.96

71.08

76.92

79.22

78.05

Clinical BERT44

72.26

74.12

73.12

79.96

80.23

80.09

Back to article page