Research Article
WSF2: A Novel Framework for Filtering Web Spam
Table 8
Sensitivity, specificity, and AUC values obtained for each classifier executed in our WSF2 platform.
| | Ratio | AUC | Sensitivity | Specificity |
| C5.0 | 1 : 17 | 0.562 | 14.2 | 98.4 | 1 : 8 | 0.649 | 50.4 | 79.5 | 1 : 4 | 0.651 | 66.4 | 65.9 | 1 : 2 | 0.648 | 77.0 | 52.7 | 1 : 1 | 0.573 | 82.3 | 32.4 |
| SVM | 1 : 17 | 0.534 | 7.1 | 99.7 | 1 : 8 | 0.590 | 26.5 | 91.6 | 1 : 4 | 0.602 | 29.2 | 91.2 | 1 : 2 | 0.604 | 30.1 | 90.8 | 1 : 1 | 0.624 | 34.5 | 90.3 |
| C5.0 + SVM | 1 : 17 | 0.579 | 17.7 | 98.1 | 1 : 8 | 0.658 | 52.2 | 79.4 | 1 : 4 | 0.713 | 63.7 | 72.9 | 1 : 2 | 0.684 | 77.0 | 52.7 | 1 : 1 | 0.646 | 34.5 | 90.2 |
| C5.0 + SVM + REGEX | 1 : 17 | 0.673 | 44.2 | 87.7 | 1 : 8 | 0.768 | 78.8 | 69.8 | 1 : 4 | 0.798 | 75.3 | 71.8 | 1 : 2 | 0.759 | 54.9 | 85.7 | 1 : 1 | 0.736 | 62.8 | 82.3 |
|
|