Skip to main content

Table 8 Results of CNN model using different pre-trained word embeddings in the first and second data sets. Bold numbers indicate the best value while underlined numbers represent the second-best value

From: Investigating the impact of pre-processing techniques and pre-trained word embeddings in detecting Arabic health information on social media

 

First data set

Second data set

Precision

Recall

F1 score

Accuracy

Precision

Recall

F1 score

Accuracy

AraVec Skip-Gram

88.16

87.87

88.01

92.7

76.84

71.96

74.32

90.6

AraVec CBOW

87.46

84.59

86

91.6

78.31

63.49

70.18

89.8

Mazajak Skip-Gram

89.04

85.25

87.1

92.3

78.26

66.67

72

90.2

Mazajak CBOW

89.44

83.28

86.25

91.9

81.05

65.61

72.51

90.6

fastText

87.46

84.59

86

91.6

84.12

56.08

67.3

89.7

ArWordVec Skip-Gram

85.76

82.95

84.33

90.6

70.56

67.2

68.83

88.5

ArWordVec CBOW

89.67

79.67

84.38

91

78.57

64.02

70.55

89.9

  1. Bold numbers indicate the best value while underlined numbers represent the second-best value