Upload README.md with huggingface_hub
Browse files
README.md
ADDED
|
@@ -0,0 +1,70 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
dataset_info:
|
| 3 |
+
features:
|
| 4 |
+
- name: "raw-request"
|
| 5 |
+
dtype: string
|
| 6 |
+
- name: "label"
|
| 7 |
+
dtype: string
|
| 8 |
+
- name: "attack_type"
|
| 9 |
+
dtype: string
|
| 10 |
+
config_name: web-attack-detection-dataset
|
| 11 |
+
---
|
| 12 |
+
|
| 13 |
+
# web-attack-detection-dataset
|
| 14 |
+
|
| 15 |
+
|
| 16 |
+
A comprehensive dataset for web attack detection and classification, containing balanced HTTP traffic data with various attack types.
|
| 17 |
+
|
| 18 |
+
This dataset includes:
|
| 19 |
+
- Normal HTTP requests for benign traffic
|
| 20 |
+
- Anomalous HTTP requests representing different attack vectors
|
| 21 |
+
- 16 different attack types including SQL injection, XSS, LFI, etc.
|
| 22 |
+
- Balanced distribution to prevent overfitting
|
| 23 |
+
|
| 24 |
+
|
| 25 |
+
## Dataset Information
|
| 26 |
+
|
| 27 |
+
|
| 28 |
+
- **Training set**: 105,757 records (11.5 MB)
|
| 29 |
+
- **Validation set**: 15,855 records (1.7 MB)
|
| 30 |
+
- **Total records**: 121,612
|
| 31 |
+
- **Attack types**: 16 categories
|
| 32 |
+
- **Format**: Apache Parquet (compressed)
|
| 33 |
+
|
| 34 |
+
|
| 35 |
+
## Data Format
|
| 36 |
+
|
| 37 |
+
- **raw-request**: Raw HTTP request string
|
| 38 |
+
- **label**: Classification label ("normal" or "anomalous")
|
| 39 |
+
- **attack_type**: Specific attack type (empty for normal requests)
|
| 40 |
+
|
| 41 |
+
## Usage
|
| 42 |
+
|
| 43 |
+
```python
|
| 44 |
+
from datasets import load_dataset
|
| 45 |
+
|
| 46 |
+
# Load the dataset
|
| 47 |
+
dataset = load_dataset("web-attack-detection-dataset")
|
| 48 |
+
|
| 49 |
+
# Access train and test splits
|
| 50 |
+
train_data = dataset['train']
|
| 51 |
+
test_data = dataset['test']
|
| 52 |
+
|
| 53 |
+
# Dataset info
|
| 54 |
+
print(f"Training samples: {len(train_data)}")
|
| 55 |
+
print(f"Test samples: {len(test_data)}")
|
| 56 |
+
```
|
| 57 |
+
|
| 58 |
+
## Statistics
|
| 59 |
+
|
| 60 |
+
|
| 61 |
+
- **Training set**: 105,757 records (11.5 MB)
|
| 62 |
+
- **Validation set**: 15,855 records (1.7 MB)
|
| 63 |
+
- **Total records**: 121,612
|
| 64 |
+
- **Attack types**: 16 categories
|
| 65 |
+
- **Format**: Apache Parquet (compressed)
|
| 66 |
+
|
| 67 |
+
|
| 68 |
+
## License
|
| 69 |
+
|
| 70 |
+
Please check the original dataset license and usage terms.
|