The process of defining correspondences between elements (fields, classes, vocabularies, constraints) of different metadata or database schemas to enable translation, integration, or interoperability of data while recording transformation rules and provenance.

Principle

Schema mapping requires semantic alignment (meaning), structural mapping (cardinality and data types), and procedural rules (transformations, value mapping, authority remapping); mappings can be lossy and must therefore record provenance and transformation logic to support interpretation and reversibility where possible.

Demonstration

Illustrative Scenario (Hypothetical): Situation — A service aggregates bibliographic records from two providers using different schemas. Recognition — The aggregator identifies that Provider A's 'creator' maps to Provider B's 'author' and that subject subdivisions in A have no direct field in B. Action — Engineers define explicit mapping rules: creator→author, subject subdivisions→combined subject + note with controlled vocabulary crosswalks; they log transformation rules and provenance. Consequence — Aggregated search returns coherent results, but some subject granularity is summarized and traceable to the original fields.

Misapplication

Assuming a universal one‑to‑one field mapping exists between schemas. The semantic error is ignoring differences in semantics, cardinality and controlled vocabularies, which leads to silent data loss or misinterpretation when fields are force‑mapped without transformation rules or provenance.

Consequence

Poor mapping produces data loss, inconsistent search and display, broken workflows, incorrect rights or embargo propagation, and increased downstream curation; well‑documented mappings facilitate quality assurance and enable informed trade‑offs.

Reversal

In federated search or minimal interoperability scenarios, intentionally coarse or partial mappings (e.g., map only title and identifier) may be preferable to preserve local semantics; conversely, ontology‑driven mappings can enhance semantic richness but require greater upfront modeling.

Boundary

Clearly within — Defining field and value correspondences between two bibliographic schemas with overlapping domains. Boundary case — Mapping between a bibliographic schema and a research data schema where semantic overlap is partial and requires conceptual modelling choices. Clearly outside — Converting between two datasets that already share an identical canonical schema (no mapping needed).

Semantic Tension

Fidelity to original semantics and granularity ↔ Practical interoperability and simplicity across heterogeneous systems.

Synthesis

Schema mapping is an explicit engineering of semantic and structural compromises; durable interoperability depends on documented transformation rules, provenance capture, and acknowledgement that some mappings are intentionally lossy to achieve broader integration.