eromang/cyberscale-contextual-training

---
language: en
license: apache-2.0
task_categories:
- text-classification
tags:
- cybersecurity
- vulnerability
- nis2
- contextual-severity
size_categories:
- 1K<n<10K
---

# CyberScale Contextual Severity Training Data

Training dataset for the CyberScale contextual severity classifier (Phase 2). Contains 32,000 scenarios combining CVE descriptions with NIS2 sector deployment contexts and cross-border exposure.

## Dataset Description

- **Source:** Parametrically generated from CVEs x 19 NIS2 sectors x cross-border conditions
- **Generation:** `training/scripts/generate_contextual.py` with `data/reference/sector_severity_rules.json`
- **Labels:** 4-class (Low=0, Medium=1, High=2, Critical=3)

## Schema

| Column | Type | Description |
|--------|------|-------------|
| `input_text` | string | Formatted input: `<description> [SEP] sector: <id> cross_border: <bool> score: <float>` |
| `label` | int | Severity class (0-3) |
| `sector` | string | NIS2 sector identifier |
| `cross_border` | bool | Cross-border exposure |
| `cvss_score` | float | CVSS v3.x base score |
| `base_severity` | string | CVSS-derived severity |
| `contextual_severity` | string | Context-adjusted severity |

## Class Distribution

| Label | Name | Count |
|-------|------|-------|
| 0 | Low | 8,000 |
| 1 | Medium | 8,000 |
| 2 | High | 8,000 |
| 3 | Critical | 8,000 |

## Citation

Part of the CyberScale project — multi-phase cyber severity assessment MCP server.