Test material
Known-answer datasets, synthetic corpora and authorized anonymized records with a documented scope.
Traceability, timestamp integrity and repeatable testing guide QUARDENCE’s engineering work.
Each stage should remain distinguishable and connected to its inputs. This makes it possible to review what the source contained, what the parser produced and what later analysis inferred.
A normalized timeline is an analytical view. It should retain the original stored value, source field, unit, known timezone or offset, conversion logic, normalized analytical timestamp and event meaning. The source timestamp remains available alongside the normalized view.
Evaluation work is planned around repeatable inputs, source-level checks and version-aware parser behavior.
Known-answer datasets, synthetic corpora and authorized anonymized records with a documented scope.
Version-aware regression tests, schema compatibility checks and cross-checks against source records.
Track false positives, false negatives, field interpretation errors and timestamp conversion errors.
Record processing versions, settings, transformation steps and known coverage gaps.
Public guidance informing the engineering approach includes:
These public references provide context for the project’s engineering direction.