Skip to content
Back to skills

Unit Detection And Normalization

ASecurity

Sub-skill of doc-extraction: Unit Detection and Normalization (+4).

  • 17 stars
  • 0 votes
  • 0 copies
  • 2 views
  • Added September 9, 2026
researchgoapi

Works with

  • api

Security analysis

A100/100

Scanned September 9, 2026

npx -y skills add vamseeachanta/workspace-hub --skill unit-detection-and-normalization --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Unit Detection And Normalization?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Unit Detection And Normalization
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/vamseeachanta-unit-detection-and-normalization/badge)](https://www.skillsdirectory.com/skills/vamseeachanta-unit-detection-and-normalization)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: doc-extraction-unit-detection-and-normalization
description: 'Sub-skill of doc-extraction: Unit Detection and Normalization (+4).'
version: 1.0.0
category: engineering
type: reference
scripts_exempt: true
---

# Unit Detection and Normalization (+4)

## Unit Detection and Normalization


Recognize units in parentheses, brackets, or inline: `(m/s²)`, `[ksi]`,
`N/mm²`. Normalize to a canonical form:

| Input variants | Canonical |
|---------------|-----------|
| `ksi`, `KSI` | `ksi` |
| `N/mm²`, `MPa` | `MPa` |
| `mA/m²`, `mA/m2` | `mA/m²` |
| `°C`, `deg C` | `°C` |
| `lb/ft`, `lbs/ft` | `lb/ft` |

Flag SI vs imperial vs field units. Prefer SI in output; keep original as
`units_original`.


## Standards Reference Parsing


Parse references like `DNV-RP-B401 Section 3.4.6` into structured form:

```yaml
standard_ref:
  body: DNV
  document: RP-B401
  edition: null         # null unless explicitly stated in source text
  edition_inferred: 2021  # set only when edition can be inferred; null otherwise
  section: 3.4.6
  table: null
  figure: null
  raw: "DNV-RP-B401 Section 3.4.6"
```

**Inference rule**: `edition` must be null unless the source text explicitly
states the year. Use `edition_inferred` only when surrounding context (title
page, header, or adjacent reference) provides strong evidence; never guess.

Common patterns:
- `DNV-RP-XXXX Sec N.N.N` / `Section N.N` / `Table N-N` / `Figure N-N`
- `API RP NNX Section N` / `API 579-1 Part N`
- `ASME BPVC Section VIII Div 2`
- `ISO NNNNN-N:YYYY Clause N.N`


## Safety and Design Factor Identification


Flag values tagged as safety factors, design factors, or usage factors:
- Keywords: "safety factor", "design factor", "utilisation factor", "usage factor"
- Often dimensionless ratios between 0 and 10
- Extract: {name, value, standard_ref, applicability}


## Material Property Recognition


Detect material properties in text or tables:
- Yield strength (SMYS), tensile strength (SMTS), Young's modulus
- Density, thermal expansion coefficient, Poisson's ratio
- Fatigue S-N curve parameters
- Extract: {property, value, units, material_grade, temperature, standard_ref}


## Condition and Applicability Tagging


Many values have applicability constraints. Tag extracted items with:
- Temperature range
- Depth/pressure range
- Material grade or category
- Environmental condition (seawater, air, buried)
- Service life assumptions

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.