Skip to main content

Advances, Systems and Applications

Table 8 A comparison evaluation of four DL models with different word embedding for AraCOVID19-SSD dataset

From: Stacked-CNN-BiLSTM-COVID: an effective stacked ensemble deep learning framework for sentiment analysis of Arabic COVID-19 tweets

DL Model

Word Embedding

Model

Precision

Recall

F-Measure

ROC

CNN

AraVec

CBOW

66.25%

65.79%

64.32%

83.55%

SG

65.18%

64.75%

58.76%

85.17%

FastText

CBOW

77.24%

71.53%

63.96%

92.76%

SG

77.93%

71.53%

64.69%

93.26%

ArWordVec

CBOW

65.37%

64.75%

62.74%

83.26%

SG

73.41%

66.82%

59.93%

86.36%

LSTM

AraVec

CBOW

77.85%

76.92%

76.5%

90.82%

SG

82.47%

82.32%

82.2%

93.91%

FastText

CBOW

86.32%

86.11%

86.03%

95.05%

SG

86.93%

86.8%

86.68%

95.9%

ArWordVec

CBOW

78.54%

78.19%

78.05%

90.31%

SG

81.16%

81.06%

80.97%

93.18%

BiLSTM

AraVec

CBOW

78.89%

78.19%

77.95%

91.87%

SG

81.93%

81.63%

81.41%

93.78%

FastText

CBOW

86.99%

87.03%

86.98%

95.54%

SG

87.16%

86.82%

86.99%

95.95%

ArWordVec

CBOW

77.96%

77.84%

77.73%

90.91%

SG

82.76%

82.66%

82.58%

93.57%

Ensemble

AraVec

CBOW

77.79%

77.15%

76.32%

90.08%

SG

84.08%

83.93%

83.87%

93.62%

FastText

CBOW

87.25%

87.26%

87.25%

97.2%

SG

86.68%

86.45%

86.29%

95.63%

ArWordVec

CBOW

76.01%

75.77%

75.4%

88.17%

SG

80.98%

80.94%

80.74%

92.39%