Busisi

Methodology

Every published link needs a source trail.

Busisi separates collection, normalization, matching, review, and publication so that a plausible relationship cannot quietly become a public claim.

01

Evidence pipeline

  1. AcquireFetch named official sources and record the retrieval context.
  2. Hash and archivePreserve an immutable copy and its cryptographic digest.
  3. Validate schemaReject malformed or incomplete records before analysis.
  4. Normalize unitsStandardize currency, dates, identifiers, and reporting scale.
  5. Reconcile controlsCompare derived totals with published control totals.
  6. Match identityLink agencies and projects using explainable evidence.
  7. Classify confidenceKeep uncertainty visible rather than forcing a match.
  8. Calculate anomaliesApply deterministic, versioned rules to reviewed data.
  9. QuarantineHold conflicts, weak matches, and failed controls out of publication.
  10. PublishExpose only records that pass the release policy.

02

Release standard

The first public register requires at least 100 reviewed linkage cases, at least 95% measured precision, at least 95% reviewer agreement, and no unresolved critical reconciliation failure.

A signal identifies a record pattern for further examination. It is not evidence of wrongdoing, waste, or unlawful conduct. Published pages will show the rule, inputs, confidence, source references, and known limitations.

Corrections will preserve prior versions and explain what changed. Records that lose their evidentiary support return to quarantine.