Expand description
Ontology alignment as optimization (CI-SKM ch 4) — given a similarity matrix
between two ontologies’ entities, find the injective correspondence set that
maximizes total alignment quality. Greedy initialization refined by
hill-climbing (reusing optimization::metaheuristics, no new optimizer).
The output is a set of graded closeMatch proposals → RequiresHumanReview
(super::correspondence); it never asserts exactMatch. Kernel-class
Divergent.
Functions§
- align
- Align two ontologies from their
n_source × n_targetsimilarity matrix. Only correspondences with degree ≥thresholdare proposed. Returns the alignment as review-requiredcloseMatchproposals.