Datasets:
metadata
task_categories:
- text-classification
language:
- en
size_categories:
- 10K<n<100K
cyber_MITRE_tactic_CTI_dataset_v16
This dataset contains procedural descriptions from the MITRE ATT&CK framework (v16, Enterprise version). Each entry includes a text column detailing a specific attack method, while the corresponding label indicates the tactic associated with the procedure. The dataset is designed for text classification tasks, aimed at identifying and categorizing attack behaviors based on the described methods.
Data Instances
An example looks as follows:
{'text': 'zwShell has established persistence by adding itself as a new service.',
'label': '['TA0003:Persistence', 'TA0004:Privilege Escalation']'}
Data Fields
The data fields are the same among all configurations:
- text (str): A procedure for an attack from MITRE.
- label (str): Corresponding technique in MITRE ATT&CK v16 (enterprise).
Curation Process
Using the Dataset You can easily download and utilize the MITRE CTI with corresponding technique with Hugging Face's datasets library:
from datasets import load_dataset
# The entire dataset is available for use
dataset = load_dataset("sarahwei/Taiwanese-Minnan-Sutiau")
print(dataset)