README.md

---
license: mit
task_categories:
  - text-classification
  - other
language:
  - en
tags:
  - cold-case
  - unsolved-murders
  - forensic-analytics
  - crime-dataset
  - pecd5
size_categories:
  - n<1K
---

# PEC5D Cold-Case Grid — dataset

Structured dataset for the PEC5D Photonic Entanglement 5D Cold-Case
Grid: 32 documented unsolved cases (1900s–present) with model-input
scores, plus entity-layer exports (characters, timelines, scenarios).

**Ethics:** hypothesis lattice only. Not legal evidence. No
accusations. Sources are public OSINT classes only.

## Files

| File | Content |
|---|---|
| `cases.json` | 32 cases: year, lat/lon, type, evidence/motive/jurisdiction/body/genetic/severity/status scores, entity tags, descriptions |
| `characters.json` | Named people (victims, suspects, POIs, witnesses) with roles and case bindings |
| `timelines.json` | Per-case dated events |
| `scenarios.json` | Hypothesis-lattice scenario variants with plausibility weights |

## Schema (cases.json)

```json
{
  "case_id": "zodiac_1968",
  "name": "The Zodiac Killer",
  "year": 1968,
  "lat": 37.7749,
  "lon": -122.4194,
  "case_type": "serial",
  "evidence": 6,        // E_i in [0,10]
  "motive": 2,          // M_i in [0,10]
  "jurisdiction": 7,    // J_i in [0,10]
  "body": 1,            // B_i in {0, 0.5, 1}
  "genetic": 0,         // G_i in {0, 1}
  "severity": 0.95,     // public impact
  "status": 0,          // S_i in {0, 0.5, 1}
  "entities": ["cipher", "letters", "..."],
  "description": "..."
}
```

## Usage

```python
from datasets import load_dataset
ds = load_dataset("Codexcoder/pec5d-cold-case-grid", split="train")
```

## Credit

Derived from publicly documented events; structured as model inputs for
the PEC5D grid (github.com/onegayunicorn/deepseek-v4-sovereign).