Datasets:

tuandunghcmut
/

cti_bench_vsp

This dataset contains 1,000 CVE descriptions with corresponding CVSS v3.1 base scores. It evaluates a model's ability to assess vulnerability severity and generate proper CVSS vector strings.

Dataset Structure

Each example contains:

url: CVE URL (typically from nvd.nist.gov)
description: CVE description detailing the vulnerability
prompt: Full instruction prompt explaining CVSS v3.1 metrics
cvss_vector: Ground truth CVSS v3.1 vector string (e.g., "CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H")
task_type: Always "vulnerability_severity_prediction"

CVSS v3.1 Metrics

The dataset covers all base metrics:

AV (Attack Vector): Network (N), Adjacent (A), Local (L), Physical (P)
AC (Attack Complexity): Low (L), High (H)
PR (Privileges Required): None (N), Low (L), High (H)
UI (User Interaction): None (N), Required (R)
S (Scope): Unchanged (U), Changed (C)
C (Confidentiality): None (N), Low (L), High (H)
I (Integrity): None (N), Low (L), High (H)
A (Availability): None (N), Low (L), High (H)

Usage

from datasets import load_dataset

# Load the dataset
dataset = load_dataset("tuandunghcmut/cti_bench_vsp")

# Access a sample
sample = dataset['train'][0]
print(f"CVE: {sample['description']}")
print(f"CVSS Vector: {sample['cvss_vector']}")

Example

CVE Description: In the Linux kernel through 6.7.1, there is a use-after-free in cec_queue_msg_fh, related to drivers/media/cec/core/cec-adap.c...

CVSS Vector: CVSS:3.1/AV:L/AC:L/PR:L/UI:N/S:U/C:N/I:N/A:H

Citation

@article{ctibench2024,
  title={CTIBench: A Benchmark for Evaluating LLMs in Cyber Threat Intelligence},
  author={[Authors]},
  journal={NeurIPS 2024},
  year={2024}
}

Original Source

This dataset is derived from CTI-Bench and is available under the same license terms.

Tasks

This dataset is designed for:

✅ Vulnerability severity assessment
✅ CVSS score calculation
✅ Risk analysis and prioritization
✅ Cybersecurity impact evaluation