Upload domains/Hranicky/only_names_and_classes/README.md with huggingface_hub
Browse files
domains/Hranicky/only_names_and_classes/README.md
ADDED
|
@@ -0,0 +1,37 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# Domain Dataset — Train/Dev/Test Splits
|
| 2 |
+
|
| 3 |
+
## Source files
|
| 4 |
+
|
| 5 |
+
| File | Records | Label |
|
| 6 |
+
|---|---|---|
|
| 7 |
+
| `benign_cesnet.json` | 461,388 | `benign_2312` |
|
| 8 |
+
| `benign_umbrella.json` | 368,956 | `benign_2310` |
|
| 9 |
+
| `malware.json` | 100,809 | `malware` |
|
| 10 |
+
| `phishing.json` | 164,425 | `phishing_since_2402`, `misp_2310` |
|
| 11 |
+
|
| 12 |
+
**Total:** 1,095,578 records
|
| 13 |
+
|
| 14 |
+
## Columns
|
| 15 |
+
|
| 16 |
+
| CSV header | Description |
|
| 17 |
+
|---|---|
|
| 18 |
+
| `input` | Domain name |
|
| 19 |
+
| `class` | `benign`, `malware`, or `phishing` |
|
| 20 |
+
| `label` | `legit` or `malicious` |
|
| 21 |
+
| `malware_type` | Malware family name (e.g. Emotet, Cobalt Strike); empty for benign/phishing |
|
| 22 |
+
|
| 23 |
+
## Split strategy
|
| 24 |
+
|
| 25 |
+
Per-class stratified split: 70% train, 15% dev, 15% test. All data used, no sampling.
|
| 26 |
+
|
| 27 |
+
## Result
|
| 28 |
+
|
| 29 |
+
Each split available as `.json` and `.csv`.
|
| 30 |
+
|
| 31 |
+
| Split | Records | Benign | Malware | Phishing |
|
| 32 |
+
|---|---|---|---|---|
|
| 33 |
+
| `train.json` / `train.csv` | 766,903 | 581,240 (75.8%) | 70,566 (9.2%) | 115,097 (15.0%) |
|
| 34 |
+
| `dev.json` / `dev.csv` | 164,335 | 124,551 (75.8%) | 15,121 (9.2%) | 24,663 (15.0%) |
|
| 35 |
+
| `test.json` / `test.csv` | 164,340 | 124,553 (75.8%) | 15,122 (9.2%) | 24,665 (15.0%) |
|
| 36 |
+
|
| 37 |
+
Generated by `create_split.py` (JSON) and `json_to_csv.py` (CSV conversion).
|