File size: 2,428 Bytes
aa80965 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 | ---
license: mit
tags:
- oil-gas
- petrochemical
- hse
- permit-to-work
- control-of-work
- simops
- risk-classification
- text-classification
- persian
- bilingual
- synthetic
- aria-ai
pretty_name: PermitGuard PTW Samples (Synthetic Bilingual)
size_categories:
- n<1K
---
# PermitGuard — Synthetic Bilingual Permit-to-Work Samples
Part of the [Aria AI](https://aria-ai.ir) oil, gas & petrochemical technical-validation portfolio
(`Aria SafeOps` → Control of Work / PTW). Companion model:
[`alirezaaminzadeh/permitguard-risk-classifier`](https://huggingface.co/alirezaaminzadeh/permitguard-risk-classifier)
and Space:
[`alirezaaminzadeh/permitguard-ptw-risk-classifier`](https://huggingface.co/spaces/alirezaaminzadeh/permitguard-ptw-risk-classifier).
## Data honesty
**This corpus is 100% synthetic.** There are no real permits, incidents, PII, contractors, or named
facilities. Equipment tags such as `T-402` / `V-101` / `P-205B` are fictional placeholders reused from
the sibling PetroSafe RAG demo unit for portfolio consistency. About **19.2%** of permits deliberately
mismatch free text vs structured flags (the text claims a precaution that the form did not actually
complete, or vice versa) — this is a designed documentation-gap feature, not a generator bug.
## Files
| File | Rows | Description |
|---|---|---|
| `permits.parquet` | 620 | Structured PTW fields + bilingual free text + rule-derived `risk_level` / `fired_rules` |
| `simops_pairs.parquet` | 260 | Balanced conflict / no-conflict pairs (5 SIMOPS rules), with JSON of both permit records |
| `eval_results.json` | — | Held-out text-classifier metrics + SIMOPS distribution |
| `benchmark_report.json` | — | Headline KPIs + methodology note |
| `manifest.json` | — | Split IDs, work types, risk levels |
**Split:** 495 train / 125 test, stratified by `risk_level`. **Non-compliant (at least one risk rule fired):** 34.2%.
## Related
- Model: [`alirezaaminzadeh/permitguard-risk-classifier`](https://huggingface.co/alirezaaminzadeh/permitguard-risk-classifier)
- Space: [`alirezaaminzadeh/permitguard-ptw-risk-classifier`](https://huggingface.co/spaces/alirezaaminzadeh/permitguard-ptw-risk-classifier)
- Architectural sibling: ContractGuard (`contractguard-clause-analyzer`)
- Product: [aria-ai.ir](https://aria-ai.ir)
MIT · Aria AI Engineering Team
|