Information governance / interactive case study
Explore how a read-only audit turns an eight-record synthetic catalog into visible evidence, duplicate candidates, and a prioritized stewardship queue—without changing a source record.
8 / 8records pass validation
75%documented provenance
75%SHA-256 coverage
7records routed for review
01 / Interactive governance review
Run the public fixture through the audit, then inspect how individual signals become a conservative, human-owned review decision.
Audit stages
Read onlyUses the same deterministic fixture results documented in the public repository.
Catalog evidence
Readysynthetic_catalog.csv / read-only
Run the audit to surface evidence gaps and duplicate candidates while keeping every catalog decision with a person.
Review decision
High priorityASSET-004 / HIGH
Confirm the source reference and identity before any reuse, relocation, merge, or deletion.
02 / Reproducible evidence
The figures below come from the repository’s included eight-record synthetic fixture. They demonstrate the audit model—not a production catalog.
The audit separates structural validity from the evidence required for confident stewardship.
Weak identity evidence and duplicate signals are prioritized without authorizing an automatic change.
Evidence, not authority
Exact names, normalized names, and checksum matches remain review candidates. They never become automatic merge, relocation, retention, or deletion decisions.
Read-only by design
The audit reports validation, coverage, provenance, confidence, and duplicate candidates, then hands each decision to a steward with the evidence still visible.
03 / Governance workflow
A repeatable sequence keeps the audit explainable and makes the next action visible without taking authority away from the people responsible for the collection.
Confirm the expected schema, controlled values, identifiers, review dates, and optional checksum format.
Calculate coverage, provenance, and confidence so evidence gaps can be compared consistently.
Surface exact-name, normalized-name, and exact-checksum candidates using conservative matching rules.
Rank records by the evidence they need and attach a clear, controlled next step.
Keep reuse, relocation, merge, retention, and deletion decisions with a human steward.
04 / Why this project matters
Digital asset inventories rarely arrive as one clean, authoritative source. They combine managed records, archive imports, department copies, incomplete provenance, and filenames that look similar without necessarily representing the same thing.
This case study shows how I approach that ambiguity: validate the structure, make evidence gaps visible, organize the work by risk, and preserve human judgment for the decisions that matter.
Scope boundary
Every figure on this page comes from the repository’s synthetic eight-record fixture. The lab stores nothing, uploads nothing, and makes no network request. Its deterministic matching is deliberately conservative; the queue organizes review and does not certify risk or authorize a catalog change.