NORM
Normalization Engines
1. Overview
Normalization Engines (NORM) are micro-engines that rescale or normalize numeric values to enable comparability across entities, time periods, or benchmarks.
NORM engines preserve the semantic meaning of a metric while expressing it relative to a declared reference base (e.g. revenue, production volume, area, FTE).
Normalization is a mathematical operation, not a representational one.
NORM is an engine type in the MICE taxonomy; concrete normalizers are implemented as MEID-registered micro engines.
2. Design Principles
-
Comparability Over Absolutes
NORM engines exist to support comparison, benchmarking, and analysis — not raw reporting. -
Explicit Reference Bases
Every normalization explicitly declares its denominator or baseline. -
Deterministic Scaling
Given identical inputs and bases, normalization always yields the same result. -
Lineage Preservation
Normalized outputs always reference their source values and bases.
3. Scope of Responsibility
3.1 What NORM engines do
NORM engines perform value rescaling without changing semantic intent:
- Convert absolute values to intensities
- Normalize metrics for peer comparison
- Express metrics relative to capacity, revenue, or activity
Typical examples:
- tCO₂e per revenue
- energy per m²
- water per unit produced
- waste per FTE
Normalization may be applied:
- after CALC,
- after AGGR,
- or at disclosure-preparation stage.
4. What NORM engines do not do
NORM engines explicitly do not perform:
- ❌ Primary computation (CALC)
- ❌ Unit or format conversion (TRFM)
- ❌ Aggregation across entities or time (AGGR)
- ❌ Metadata tagging or classification (META)
- ❌ Estimation or extrapolation (SEM)
- ❌ Policy or routing decisions (ZSSR)
Normalization changes scale, not structure, meaning, or policy.
5. Inputs
NORM engines consume:
- Validated numeric values (absolute or aggregated)
- A declared normalization base, such as:
- revenue
- production volume
- floor area
- headcount
- Optional benchmark or peer group context
Inputs must be numerically valid and temporally aligned.
6. Outputs
NORM engines produce:
- Normalized value payload
- Explicit reference metadata, including:
- base type
- base value
- normalization rule identifier
- Provenance links to source signals and base signals
Normalized outputs are typically consumed by:
- reporting and disclosure pipelines,
- benchmarking and analytics,
- scenario and scoring engines.
7. Position in the computation chain
Normalization typically occurs after computation and aggregation:
INPUT
→ VALI
→ CALC
→ TRFM
→ AGGR
→ NORM
→ META
→ Reporting / Analysis
This ordering ensures that normalization operates on trusted, consolidated values.
8. Canonical identification
- Engine Type: NORM
- USO role: expresses values relative to a declared base
- Category: Micro Engine (MICE)
NORM identifiers appear in:
- signal lineage (CMI stamps),
- audit trails,
- explainability layers.
9. Registry view
All engines tagged as norm:
10. Design rationale
Normalization is not a formatting concern — it is a semantic scaling operation.
By isolating normalization into its own engine type, ZAYAZ ensures:
- clear separation between absolute truth and relative insight,
- auditability of intensity metrics,
- consistent treatment of denominators across reports,
- safe reuse of normalization logic across domains.
11. Related engine types
- CALC — Calculation Engines
- AGGR — Aggregation Engines
- TRFM — Transformation Engines
- META — Metadata Enrichment Engines
- SCEN — Scenario Engines
Stable