A technique for resolving semantic heterogeneity by translating, reconciling, or otherwise transforming differing conceptualizations between data sources—typically via an intermediary model, mapping rules, or runtime translators—so that heterogeneous systems can exchange, query, or aggregate information meaningfully.

Principle

Mediation uses an explicit intermediary (mediator ontology, mapping functions, or translation rules) to map source-specific concepts and representations to a common or compatible form; interoperability is achieved by applying these transformations at query or integration time.

Demonstration

Illustrative scenario → Two environmental datasets report temperature: one records 'Temp_Celsius' as numeric Celsius, the other 'AirTemp' tagged with units 'F'. Recognition → mediator identifies differing units and semantically equivalent concept. Action → mediation applies unit conversion rule and aligns variable names to a shared mediator schema. Consequence → integrated analysis computes trends across both datasets with consistent units.

Misapplication

Assuming mediation can reconcile contradictory ontological commitments (e.g., fundamentally different scopes or intended referents) without information loss; treating mediation as a transparent, lossless operation leads to incorrect merges or misleading summaries.

Consequence

Effective mediation enables cross-source queries, fused analytics, and user‑level interoperability; imperfect mediation can introduce ambiguity, drop distinctions, or produce misleadingly homogenized results.

Reversal

Mediation is inappropriate when provenance, auditability, or legal/ethical constraints require preserving original conceptualizations intact, or when sources are purposely semantically incompatible; in such cases, negotiated interfaces, side-by-side presentation, or proxy queries may be used instead.

Boundary

Clearly within: applying mediator mappings and transformations to reconcile differing schemas and semantics at integration/query time. Boundary case: using a lookup table of equivalent labels without structural transformation. Clearly outside: syntactic translation (e.g., character encoding conversion) that does not alter conceptual semantics.

Semantic Tension

Fidelity versus interoperability — preserving source-specific distinctions retains accuracy but limits integration; enforcing uniform semantics increases usability but risks semantic loss.

Synthesis

Semantic mediation is a pragmatic reconciliation strategy: it sacrifices some local specificity when necessary to produce useful, interpretable integrated views, and must explicitly record transformations and residual uncertainties.