Research Article
An Improved Feature Extraction Approach for Web Anomaly Detection Based on Semantic Structure
Table 2
The performance comparison of difference models.
| Algorithm | Precision | Recall | F1-score | Accuracy |
| Random forest | 0.8899 | 0.7860 | 0.8348 | 0.8169 | Random forest | 0.9154 | 0.8508 | 0.8819 | 0.8376 | Decision tree | 0.8646 | 0.8075 | 0.8351 | 0.8124 | Decision tree | 0.9169 | 0.8454 | 0.8797 | 0.8352 | Support vector machine | 0.6746 | 0.9090 | 0.7745 | 0.6912 | Support vector machine | 0.9636 | 0.8541 | 0.9056 | 0.8731 | K-neighbours | 0.8879 | 0.7764 | 0.8292 | 0.8109 | K-neighbours | 0.9335 | 0.8491 | 0.8893 | 0.8493 |
|
|
The algorithms with are trained with semantic structure, and others are trained with conventional routine. ā |