Skip to main content

Table 5 Performance measures (%) for different feature sets and different classification techniques

From: Comparing writing style feature-based classification methods for estimating user reputations in social media

Feature set

Reputation s

Base learners

C4.5

NN

SVM

NB

Precision

Recall

F-measure

Precision

Recall

F-measure

Precision

Recall

F-measure

Precision

Recall

F-measure

(a) Segmenting type s = like

 F1

Good

63.89

46.00

53.49

56.76

63.00

59.72

64.38

47.00

54.34

64.38

47.00

54.34

Bad

57.81

74.00

64.91

58.43

52.00

55.03

58.27

74.00

65.20

58.27

74.00

65.20

 F1 + F2

Good

60.94

39.00

47.56

54.55

54.00

54.27

63.08

41.00

49.70

59.38

19.00

28.79

Bad

55.15

75.00

63.56

54.46

55.00

54.73

56.30

76.00

64.68

51.79

87.00

64.93

 F1 + F2 + F3

Good

60.94

39.00

47.56

54.64

53.00

53.81

63.08

41.00

49.70

58.06

18.00

27.48

Bad

55.15

75.00

63.56

54.37

56.00

55.17

56.30

76.00

64.68

51.48

87.00

64.68

 F1 + F2 + F3 + F4

Good

65.38

51.00

57.30

69.05

58.00

63.04

77.94

53.00

63.10

83.33

50.00

62.50

Bad

59.84

73.00

65.77

63.79

74.00

68.52

64.39

85.00

73.28

64.29

81.00

71.68

(b) Segmenting type s = dislike

 F1

Good

58.62

34.00

43.04

67.50

27.00

38.57

49.33

37.00

42.29

63.41

26.00

36.88

Bad

53.52

76.00

62.81

54.38

87.00

66.92

49.60

62.00

55.11

53.97

68.00

60.18

 F1 + F2

Good

59.34

54.00

56.54

64.06

41.00

50.00

59.74

46.00

51.98

64.58

31.00

41.89

Bad

57.80

63.00

60.29

56.62

77.00

65.25

56.10

69.00

61.88

56.67

17.00

26.15

 F1 + F2 + F3

Good

59.34

54.00

56.54

64.06

41.00

50.00

59.74

46.00

51.98

64.58

31.00

41.89

Bad

57.80

63.00

60.29

56.62

77.00

65.25

56.10

69.00

61.88

56.67

17.00

26.15

 F1 + F2 + F3 + F4

Good

70.00

63.00

66.32

73.63

67.00

70.16

77.91

67.00

72.04

73.63

67.00

70.16

Bad

66.36

73.00

69.52

69.72

76.00

72.73

71.05

81.00

75.70

77.11

64.00

69.95

(c) Segmenting type s = sum

 F1

Good

65.33

49.00

56.00

57.61

53.00

55.21

61.64

45.00

52.02

95.24

20.00

33.06

Bad

59.20

74.00

65.78

56.48

61.00

58.65

56.69

72.00

63.44

55.06

98.00

70.50

 F1 + F2

Good

64.71

55.00

59.46

58.95

56.00

57.44

69.14

56.00

61.88

75.47

40.00

52.29

Bad

60.87

70.00

65.12

58.10

61.00

59.51

63.03

75.00

68.49

59.18

87.00

70.45

 F1 + F2 + F3

Good

64.71

55.00

59.46

60.64

57.00

58.76

69.14

56.00

61.88

75.47

40.00

52.29

Bad

60.87

70.00

65.12

59.43

63.00

61.17

63.03

75.00

68.49

59.18

87.00

70.45

 F1 + F2 + F3 + F4

Good

69.12

47.00

55.95

58.62

51.00

54.55

70.11

61.00

65.24

75.68

56.00

64.37

Bad

59.85

79.00

68.10

56.64

64.00

60.09

65.49

74.00

69.48

65.08

82.00

72.57

(d) Segmenting type s = portfolio

 F1

Good

60.00

57.00

58.46

59.62

62.00

60.78

61.11

55.00

57.89

54.93

78.00

64.46

Bad

59.05

62.00

60.49

60.42

58.00

59.18

59.09

65.00

61.90

62.07

36.00

45.57

 F1 + F2

Good

65.66

65.00

65.33

62.63

62.00

62.31

63.37

64.00

63.68

56.94

82.00

67.21

Bad

65.35

66.00

65.67

62.38

63.00

62.69

63.64

63.00

63.32

67.86

38.00

48.72

 F1 + F2 + F3

Good

65.66

65.00

65.33

56.03

65.00

60.19

63.37

64.00

63.68

56.94

82.00

67.21

Bad

65.35

66.00

65.67

58.33

49.00

53.26

63.64

63.00

63.32

67.86

38.00

48.72

 F1 + F2 + F3 + F4

Good

62.83

71.00

66.67

69.16

74.00

71.50

69.91

79.00

74.18

76.77

76.00

76.38

Bad

66.67

58.00

62.03

72.04

67.00

69.43

75.86

66.00

70.59

76.24

77.00

76.62

Feature set

Reputation s

Ensemble learning methods

RS-C4.5

RS-NN

RS-SVM

RS-NB

Precision

Recall

F-measure

Precision

Recall

F-measure

Precision

Recall

F-measure

Precision

Recall

F-measure

(a) Segmenting type s = like

 F1

Good

66.28

57.00

61.29

64.36

65.00

64.68

67.42

60.00

63.49

66.67

58.00

62.03

Bad

62.28

71.00

66.36

64.65

64.00

64.32

63.96

71.00

67.30

62.83

71.00

66.67

 F1 + F2

Good

70.45

62.00

65.96

71.76

61.00

65.95

71.26

62.00

66.31

88.00

22.00

35.20

Bad

66.07

74.00

69.81

66.09

76.00

70.70

66.37

75.00

70.42

55.43

97.00

70.55

 F1 + F2 + F3

Good

70.45

62.00

65.96

71.43

65.00

68.06

71.91

64.00

67.72

86.96

20.00

32.52

Bad

66.07

74.00

69.81

67.89

74.00

70.81

67.57

75.00

71.09

54.80

97.00

70.04

 F1 + F2 + F3 + F4

Good

81.40

70.00

75.27

83.72

72.00

77.42

96.74

89.00

92.71

96.34

79.00

86.81

Bad

73.68

84.00

78.50

75.44

86.00

80.37

89.81

97.00

93.27

82.57

90.00

86.12

(b) Segmenting type s = dislike

 F1

Good

54.76

69.00

61.06

52.26

81.00

63.53

52.25

93.00

66.91

71.88

23.00

34.85

Bad

58.11

43.00

49.43

57.78

26.00

35.86

68.18

15.00

24.59

60.71

34.00

43.59

 F1 + F2

Good

62.96

68.00

65.38

63.93

78.00

70.27

61.83

81.00

70.13

71.62

53.00

60.92

Bad

65.22

60.00

62.50

71.79

56.00

62.92

72.46

50.00

59.17

87.80

36.00

51.06

 F1 + F2 + F3

Good

63.11

65.00

64.04

66.09

76.00

70.70

61.54

80.00

69.57

71.62

53.00

60.92

Bad

63.92

62.00

62.94

71.76

61.00

65.95

71.43

50.00

58.82

87.80

36.00

51.06

 F1 + F2 + F3 + F4

Good

88.30

83.00

85.57

98.88

88.00

93.12

94.90

93.00

93.94

86.79

92.00

89.32

Bad

83.96

89.00

86.41

89.19

99.00

93.84

93.14

95.00

94.06

100.00

79.00

88.27

(c) Segmenting type s = sum

 F1

Good

73.68

70.00

71.79

69.15

65.00

67.01

66.67

52.00

58.43

82.93

34.00

48.23

Bad

71.43

75.00

73.17

66.98

71.00

68.93

60.66

74.00

66.67

58.49

93.00

71.81

 F1 + F2

Good

78.16

68.00

72.73

63.77

44.00

52.07

73.91

68.00

70.83

77.22

61.00

68.16

Bad

71.68

81.00

76.06

57.25

75.00

64.94

70.37

76.00

73.08

67.77

82.00

74.21

 F1 + F2 + F3

Good

75.58

65.00

69.89

64.06

41.00

50.00

73.03

65.00

68.78

77.22

61.00

68.16

Bad

69.30

79.00

73.83

56.62

77.00

65.25

68.47

76.00

72.04

67.77

82.00

74.21

 F1 + F2 + F3 + F4

Good

82.72

67.00

74.03

50.00

100.00

66.67

81.55

84.00

82.76

83.87

78.00

80.83

Bad

72.27

86.00

78.54

0.00

0.00

0.00

83.51

81.00

82.23

80.20

81.00

80.60

(d) Segmenting type s = portfolio

 F1

Good

58.82

60.00

59.41

57.29

55.00

56.12

60.95

64.00

62.44

65.09

69.00

66.99

Bad

59.18

58.00

58.59

56.73

59.00

57.84

62.11

59.00

60.51

67.02

63.00

64.95

 F1 + F2

Good

70.09

75.00

72.46

64.15

68.00

66.02

76.19

80.00

78.05

71.43

80.00

75.47

Bad

73.12

68.00

70.47

65.96

62.00

63.92

78.95

75.00

76.92

77.27

68.00

72.34

 F1 + F2 + F3

Good

71.56

78.00

74.64

69.70

69.00

69.35

71.13

69.00

70.05

71.93

82.00

76.64

Bad

75.82

69.00

72.25

69.31

70.00

69.65

69.90

72.00

70.94

79.07

68.00

73.12

 F1 + F2 + F3 + F4

Good

77.78

84.00

80.77

84.26

91.00

87.50

94.06

95.00

94.53

85.98

92.00

88.89

Bad

82.61

76.00

79.17

90.22

83.00

86.46

94.95

94.00

94.47

91.40

85.00

88.08

  1. For each segmenting type, the best results with respect to the three performance measures are highlighted in italics, and the best precision, recall, and F-measure over all 128(=4 feature types × 8 classification techniques × 4 segmenting types) are additionally highlighted as bold italics