Datasets:
metadata
license: cc0-1.0
language:
- en
task_categories:
- text-classification
tags:
- cyber-security
- vulnerability
- cve
- cwe
- cvss
- nvd
- tabular
- time-series
- risk-analysis
pretty_name: NVD CVE & CWE Vulnerability Dataset (1999-2025)
size_categories:
- 100K<n<1M
This collection brings together every Common Vulnerabilities & Exposures (CVE) entry published in the National Vulnerability Database (NVD) from the very first identifier — CVE-1999-0001 — through all records available on 30 May 2025.
It was built automatically with a Python script that calls the NVD REST API v2.0 page-by-page, handles rate-limits, and filters data.
After download each CVE object is pared down to the essentials and written to CVE_CWE_2025.csv with the following columns:
| Column | Meaning |
|---|---|
ID |
Simple row index (starting at 1) |
CVE-ID |
Official CVE identifier |
CVSS-V4, CVSS-V3, CVSS-V2 |
Base scores for the three CVSS generations; missing versions are stored as the string "None" |
SEVERITY |
Qualitative severity (e.g. LOW, MEDIUM, HIGH, CRITICAL) taken from the newest CVSS version present |
DESCRIPTION |
Normalised English description from the NVD feed |
CWE-ID |
Mapped Common Weakness Enumeration code (CWE-###) or NVD-CWE-Other when no specific mapping exists |
Possible uses
- Exploratory analysis of vulnerability trends over 25 years (e.g. growth of certain CWE classes or shifts in average CVSS score).
- Machine-learning tasks such as severity or CWE prediction from free-text descriptions.
- Joining with product-specific SBOMs or asset inventories to assess organisational exposure.
- Quick look-ups for incident response, education, security research, or reporting dashboards.