Skip to main content
Jira progress: loading…

RULES

Rule-Sets, Signal Governance & Enforcement Logic in ZAYAZ

Design principle: Precision before automation. Rule-sets in ZAYAZ exist to protect compliance meaning, data trust, and auditability — not to micromanage every field.


1. Purpose of Rule-Sets in ZAYAZ

Rule-sets in ZAYAZ are governance instruments, not UI constraints.

They exist to ensure that when a signal:

  • affects compliance obligations,
  • participates in computation or inference,
  • or introduces trust or assurance risk,

…the platform can enforce correctness, explainability, and traceability in a deterministic way.

ZAYAZ intentionally treats every column in every table as a signal within the Smart Searchable Signal Registry (SSSR). However, not every signal requires a rule-set.

This chapter defines:

  • when a signal requires rule governance,
  • how governance levels are applied,
  • and how signals are automatically classified at scale.

2. Core Design Decision: “Everything is a Signal”

In ZAYAZ, every column is registered in SSSR as a signal — including:

  • identifiers,
  • system metadata,
  • descriptions,
  • operational fields,
  • and compliance-critical metrics.

This design enables:

  • uniform AI interpretation (ZARA / ZAAM),
  • cross-table reasoning,
  • future-proof extensibility.

However:

Rule-sets are applied selectively — not universally.

The system must therefore distinguish between:

  • governed signals (compliance-relevant), and
  • non-governed signals (metadata / operational).

3. When Does a Signal Require a Rule-Set?

ZAYAZ uses three binary tests. A signal requires a rule-set if any one of them evaluates to true.

Rule-Set Requirement Decision

rule_required = Disclosure_Relevance
OR Computation_Relevance
OR Trust_Risk

3.1. Test A — Disclosure Relevance (Framework-Linked)

A signal is governed if it is referenced by any disclosure framework, including but not limited to:

  • CSRD / ESRS
  • EU Taxonomy
  • GRI
  • XBRL / iXBRL mappings
  • National or sector frameworks

Examples

  • Total energy consumption (kWh)
  • Scope 1 emissions
  • Workforce headcount

Implementation

  • Determined via framework_signal_map
  • No hard-coding inside rules

3.2. Test B — Computation Relevance (Engine-Linked)

A signal is governed if it is:

  • an input to a Micro Engine (MICE), or
  • an output of a Micro Engine, or
  • part of derived or aggregated metrics.

This includes engines in categories such as:

  • CALC (calculation)
  • VALI (validation)
  • NORM (normalization)
  • AGGR (aggregation)
  • SEM (estimation / extrapolation)

Rationale If a signal participates in computation, its correctness and provenance must be enforced.


3.3. Test C — Trust Risk (Human or External Origin)

A signal is governed if it is:

  • entered manually by a human,
  • imported from external systems (ERP, API, OCR),
  • or used as evidence, policy, or assertion requiring assurance.

Rationale Any signal that can introduce error, bias, or misrepresentation must be rule-controlled.


4. Signal Kind Classification (signal_kind)

To prevent rule inflation, every signal is classified by signal kind.

This is orthogonal to governance level and enables cheap, deterministic filtering.

Signal Kind Enum (Minimal Canonical Set)

signal_kindDescriptionDefault Rule Requirement
identifierIDs, keys, ECO numbers, UUIDs🔴 No
system_metacreated_at, row_version, source_system🔴 No
descriptive_metanames, descriptions, notes🔴 No
control_metaworkflow state, status flags🔴 No
measurequantitative values with units⚠️ Evaluate A/B/C
attributecategorical facts (country, NACE)⚠️ Evaluate A/B/C
document_evidencefiles, URLs, certificates✅ Yes
computedderived outputs✅ Yes

Important: identifier, system_meta, and descriptive_meta signals are still signals — they simply do not require rule-sets beyond schema constraints.


5. Governance Level Model (Compact & Deterministic)

Instead of one-off rule logic, ZAYAZ applies governance levels.

Each governed signal is assigned exactly one level.

Governance Levels

LevelNamePurpose
0noneNo rule-set; schema only
1basicFormat, unit, range checks
2complianceCompleteness, applicability, framework logic
3assuranceEvidence, provenance, verifier readiness

5.1. Governance Level Semantics

Level 0 — none

  • Identifiers
  • System metadata
  • Descriptive text

Controls

  • Datatype
  • Length
  • FK / uniqueness

Level 1 — basic Used for operational but non-reportable signals.

Controls

  • Datatype & unit enforcement
  • Range checks
  • Enum constraints

Level 2 — compliance Used for framework-relevant signals.

Controls

  • Applicability rules (NACE, size, geography)
  • Completeness rules
  • Cross-signal consistency
  • Framework mapping validation

Level 3 — assurance Used for signals that may be audited or verified.

Controls

  • Evidence requirements
  • Source provenance
  • Trust scoring (DaVE / VTE)
  • Verifier workflow integration
  • Audit logging (ALTD)

6. Auto-Classifier Specification (SSSR at Scale)

To avoid manual tagging, ZAYAZ includes an auto-classification mechanism implemented as a TAGG micro-engine.

6.1. Heuristic Classification Rules

Identifier Detection

  • Column name ends with _id, _uuid, _key, _ref
  • Datatype: UUID, PK, FK

signal_kind = identifier


System Metadata Detection

  • Names like created_at, updated_at, ingested_at
  • Source tracking fields

signal_kind = system_meta


Descriptive Metadata Detection

  • Names containing description, comment, note, label

signal_kind = descriptive_meta


Measure Detection

  • Numeric datatype
  • Unit defined or expected
  • Time granularity present

signal_kind = measure


Evidence Detection

  • Names containing file, url, certificate, attachment
  • Datatype: file reference, blob, URL

signal_kind = document_evidence


6.2. Governance Level Assignment Logic

if signal_kind in [identifier, system_meta, descriptive_meta]:
governance_level = 0
else if Disclosure_Relevance OR Computation_Relevance:
governance_level = 2
else if Trust_Risk:
governance_level = 1
if document_evidence OR verifier_required:
governance_level = 3

This logic is:

  • deterministic,
  • explainable,
  • auditable,
  • overrideable by Super Admins.

7. Relationship to Rule Primitives (Forward Reference)

This chapter defines when rule-sets apply and at what level.

The Rule Primitive Taxonomy (defined in the next chapter) defines:

  • how rules are composed,
  • what rule primitives exist (applicability, validation, computation),
  • where rules are executed (engines, not forms).

Key separation: Governance Level decides if rules apply. Rule Primitives decide how they apply.


8. Architectural Guarantees

This model guarantees that ZAYAZ:

  • avoids rule explosion,
  • scales across frameworks,
  • preserves AI interpretability,
  • remains audit-ready,
  • and survives regulatory change.

ZAYAZ governs reality — not checklists. Frameworks are mappings. Signals are truth. Rules protect meaning.




GitHub RepoRequest for Change (RFC)