Table 13 Comparative test results.

From: An emotion-oriented product evaluation model based on BP neural network and improved dempster–shafer theory

Sample number

Fusion method

Accuracy (Acc, %)

Conflict degree (K)

Stability (S)

Sample 2

Traditional D-S

73.3

0.182

0.071

Murphy averaging

76.0

0.138

0.064

Yager rule

78.7

0.000

0.076

Proposed method

86.7

0.056

0.041

Sample 9

Traditional D-S

70.0

0.205

0.076

Murphy averaging

74.7

0.163

0.068

Yager rule

77.3

0.000

0.079

Proposed method

85.3

0.049

0.044

Sample 14

Traditional D-S

75.3

0.192

0.069

Murphy averaging

78.0

0.146

0.057

Yager rule

79.3

0.000

0.075

Proposed method

87.3

0.061

0.040

Sample 20

Traditional D-S

72.0

0.186

0.075

Murphy averaging

76.7

0.151

0.061

Yager rule

78.0

0.000

0.078

Proposed method

88.0

0.053

0.042

  1. Notes:
  2. Accuracy (Acc, %): Proportion of fusion results matching expert consensus categories.
  3. Conflict Degree (K): Degree of inconsistency during evidence fusion.
  4. Stability (S): Output standard deviation under ± 0.5 noise perturbation in expert sco.
Back to article page