dataeval.core.LabelAlignmentResult¶

class dataeval.core.LabelAlignmentResult¶

Result of aligning a source vocabulary against a target ontology.

correspondences¶

Accepted correspondences, all relations: equivalent (exact or matcher), narrower (structurally entailed coarsening), and broader (granularity-mismatch diagnostics). A custom matcher may also contribute related.

Type:¶

Sequence[Correspondence]

unaligned_source¶

Source concept ids with no carryable correspondence (not in class_remap). Read open-world: out-of-vocabulary with respect to the target, not invalid.

Type:¶

Sequence[str]

unaligned_target¶

Target concept ids never referenced by any correspondence — the part of the reference vocabulary the source does not cover.

Type:¶

Sequence[str]

class_remap¶

source id -> target id for the correspondences that license a rewrite (equivalent and narrower only): the safe carry-over of source labels into the target vocabulary. broader/related are excluded.

Type:¶

Mapping[str, str]

mergeability¶

How completely the source is expressible in the target. "lossless" — every source concept is carryable and class_remap is injective (no two sources collapse to one target). "lossy" — every source is carryable but class_remap collapses two or more sources to a single target (specificity lost). "partial" — at least one source concept is unaligned (or only broader/related) and cannot be carried over. Generalizes reconciliation conformance (conformant = lossless, all equivalent).

Type:¶

{“lossless”, “lossy”, “partial”}