jvaquet/cveinfo-2025-11-21

---
dataset_info:
features:
- name: id
dtype: string
- name: text
dtype: string
- name: entity_spans
list:
- name: label
dtype: string
- name: span
list: int64
- name: text
dtype: string
splits:
- name: train
num_bytes: 17802738
num_examples: 27161
- name: validation
num_bytes: 4411069
num_examples: 6790
- name: test
num_bytes: 4442634
num_examples: 6790
download_size: 8048607
dataset_size: 26656441
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: validation
path: data/validation-*
- split: test
path: data/test-*
---

# Overview
- Dataset to extract from CVE vulnerability descriptions:
- Product
- Vendor
- Version (in different semantics: equal, less_than, less_than_or_equal)
- Original data is taken from github [`CVEProject/cvelistV5`](https://github.com/CVEProject/cvelistV5/tree/f884daec923e821fcbbdaade7aa6345d4b9cae01) on 2025-11-21.
- The data is preprocessed as follows:
- Filtering to only contain datapoints with ground truth labels.
- CVE Title and Description are concatenated.
- For each ground truth label, the spans are extracted.