Research Article
WSF2: A Novel Framework for Filtering Web Spam
Table 7
AUC results for C5.0 and SVM classifiers working together with regular expressions.
| | Class-imbalance ratio | | 1 : 17 | 1 : 8 | 1 : 4 | 1 : 2 | 1 : 1 |
| C5.0 | 0.562 | 0.649 | 0.651 | 0.648 | 0.573 | SVM | 0.534 | 0.590 | 0.602 | 0.604 | 0.624 | C5.0 + SVM | 0.579 | 0.658 | 0.713 | 0.684 | 0.646 | C5.0 + SVM + REGEX | 0.673 | 0.768 | 0.798 | 0.759 | 0.736 |
|
|