Interpretation integrity audit
Canonical definition. This page fixes the operational meaning of interpretation integrity audit within the doctrine of interpretive governance. It defines the object, the perimeter, and the limits of the audit without disclosing internal thresholds, weighting schemes, annotation procedures, or execution templates.
An interpretation integrity audit is a formal procedure that produces an opposable report qualifying the gap between a frozen canonical corpus (snapshot), outputs produced under declared conditions, and a traceable evidence chain linking each finding to verifiable artifacts.
Identifier and synonyms
- Term: Interpretation integrity audit
- Alternative term: Interpretive audit (generic usage, to avoid when integrity is the measured object)
- Measured object: interpretive integrity: anchoring, distortion, stability, and traceability
Object
The audit makes the following elements verifiable: the assertions produced by a system about a given corpus; the status of those assertions (anchored, inferred, contradictory, undetermined); the presence or absence of opposable evidence; and the declared conditions under which the outputs were obtained. A score is not an audit: the audit produces a structured artifact and an evidence chain, whereas a score is a derived projection of that artifact.
Primary unit
- A bounded audit perimeter: domain, subdirectory, URL list, or corpus list.
- A timestamp and a frozen snapshot of the audited corpus.
- A cryptographic fingerprint of the snapshot (for example SHA-256).
- A declared corpus-construction method: crawl, export, or enumerated corpus.
Audit artifact
- audit metadata (identifier, date, protocol, issuer);
- the audit unit (snapshot and perimeter);
- a versioned QuerySet;
- runs and raw outputs;
- typed findings linked to evidence;
- a condition-based validity statement.
Conditional validity
The validity of an interpretation integrity audit is conditional, not purely calendar-based. It remains valid only as long as the critical audit variables do not change: snapshot hash, perimeter, QuerySet version, declared model identity, declared execution parameters, and run timestamps. Once one of those variables changes, the audit must be refreshed.
Relation to IIP-Scoring
IIP-Scoring™ is a projection function applied to audit results in order to expose comparable metrics, and when defined, gates or grades. Scoring summarizes; it never replaces the audit report.
Relation to interpretive auditability
Interpretive auditability is a property: it describes the minimum conditions that make an interpretation explainable, verifiable, and contestable. The interpretation integrity audit is the procedure that exploits that property in order to produce an opposable report.
Non-goals
- It is not a legal opinion or a compliance certification.
- It does not guarantee the absence of hallucination.
- It does not measure traffic, SEO performance, or conversion.
- It does not disclose internal scoring thresholds or annotation procedures.
Why this term matters
This term names the full audit process rather than a single test or a single metric. It links canon definition, authority boundary, evidence gathering, gap diagnosis, correction, and monitoring. Without that broader sense, audit risks being reduced to a cosmetic score.
Why this is broader than a score
The audit can contain scores, but it is never reducible to a score. It is a governance process that connects interpretation, evidence, correction, and monitoring in one bounded cycle.