Semantic Interoperability & Ontological Layering

• Semantic interoperability and ontological layering are defined as the ability of independent systems to exchange and accurately interpret data through modular, hierarchical ontologies that standardize concepts and mappings.
• Ontological layering structures models into tiers—from abstract top ontologies to specific application layers—facilitating explicit mappings and reducing semantic drift across domains.
• Practical implementations in fields such as astronomy, materials science, and healthcare illustrate how layered architectures enhance automated reasoning, schema alignment, and system interoperability.

Semantic interoperability denotes the ability of independent systems, models, or agents to exchange, interpret, and act upon information in a way that preserves intended meaning across technical and organizational boundaries. Ontological layering is the core architectural discipline that enables and sustains semantic interoperability at scale: ontologies are modularized and stacked into hierarchies or tiers, each with well-defined scope and mapping relations, ensuring that concepts, relations, and operational mechanisms at each layer align across domains, tools, or communities of practice (Beverley et al., 14 Jun 2025, Vogt et al., 6 May 2024, Pessemier et al., 2013, Vogt, 2023, Kent, 2018). This entry provides a comprehensive account of semantic interoperability through the lens of ontological layering, covering its formalization, layering topologies and mechanisms, methodological patterns, representative architectures, and implementation exemplars.

1. Formal Foundations and Definitions

Semantic interoperability is more than data exchange; it requires that exchanged information is interpretable in congruence with the sender’s intended semantics across heterogeneous systems. Ontologies—formal, explicit specifications of shared conceptualizations—provide the backbone for this process (Beverley et al., 14 Jun 2025, Vogt et al., 6 May 2024). The contemporary formalization distinguishes several intertwined axes and subtypes:

• Human–Human Interoperability (HHI): Shared conceptual understanding among experts, evaluated by the clarity and unambiguity with which naturally formulated competency questions can be answered across communities (Beverley et al., 14 Jun 2025).
• Human–Machine Interoperability (HMI): The facility for a machine to faithfully interpret formalized human knowledge, typically captured in OWL2, RDF(S), or DL-based ontologies and operationalized via decidability and consistency guarantees.
• Machine–Machine Interoperability (MMI): Unattended semantic coherence between autonomous computational systems, often mediated by interface contracts between ontologies and realized via federated reasoning across mapped schemas (Beverley et al., 14 Jun 2025, Dunbar et al., 2022).

Within this, "semantic interoperability" is decomposed into (Vogt et al., 6 May 2024):

• Terminological Interoperability: Alignment of terms by meaning (ontological, via intensional definitions) and by referent (referential, via extension or instance alignment).
• Propositional Interoperability: Consistency in structuring and interpreting statements, encompassing schema (structural) and logical interoperability.

Semantically robust systems must exhibit all these properties, at both static (design-time) and dynamic (run-time) operational stages (Pessemier et al., 2013).

2. Ontological Layering: Architectures and Types

Ontological layering is the systematic decomposition of conceptual models into stacked, modular tiers, each handling a distinct level of abstraction or scope. The dominant patterns are (Beverley et al., 14 Jun 2025, Dunbar et al., 2022, Vogt et al., 6 May 2024, Rovetto, 2017, Pessemier et al., 2013):

1. Top/Upper Ontology: Domain-neutral, philosophically grounded primitives (e.g., Object, Process, Event) that act as abstract anchors. Examples include BFO, DOLCE, SUMO, EMMO (Dunbar et al., 2022, Rovetto, 2017, Horsch et al., 2020, Horsch et al., 2020).
2. Mid-Level/Core Ontology: Cross-cutting patterns (e.g., roles, participation, n-ary relationships, mereotopological structure). These synthesize patterns applicable across domains and support interoperability between domain ontologies (Beverley et al., 14 Jun 2025, Dunbar et al., 2022, Horsch et al., 2020).
3. Domain/Reference Ontology: Encodes subject-matter concepts, relationships, rules, and constraints for a particular sector (e.g., Cybersecurity, Orbital Debris, Risk Management).
4. Application/Model Layer: Describes concrete systems or datasets (e.g., a telescope instrument, an EHR database), formulated as instances and specific configurations using domain-layer vocabulary (Pessemier et al., 2013, Berges et al., 22 Jan 2024).
5. Instance/Data Layer: Populated with specific objects, facts, and events; referenced and structured by the layers above.

Transitions between layers typically involve explicit subsumption, equivalence, or mapping rules. For example, in the DEFII framework, each L₂ (domain) class must be an rdfs:subClassOf some L₁ (core) class (Dunbar et al., 2022).

Layered alignment is not only conceptual but is enforced operationally: ontological imports (OWL owl:imports), explicit mapping functions (SPARQL CONSTRUCT, SWRL rules), and namespace hierarchies operationalize separation-of-concerns and mitigate semantic drift (Horsch et al., 2020, Rovetto, 2017, Berges et al., 22 Jan 2024).

3. Mapping, Alignment, and Interoperability Mechanisms

Semantic interoperability in layered architectures is realized through formal mapping and alignment techniques operating at each layer:

• Terminological mappings: Ontological (intensional) and referential (extensional) mappings, formalized as $M_{ont} \subset V_1 \times V_2$ with $(t_1, t_2) \in M_{ont}$ when $$Intension_{\mathcal{O}_1}(t_1) = Intension_{\mathcal{O}_2}(t_2)$$, and $M_{ref}$ for extensions (Vogt et al., 6 May 2024). These are instantiated in practice as owl:equivalentClass, owl:equivalentProperty, or bespoke mapping properties.
• Schema crosswalks: Schema-level slot alignments $X_{sch} \subset Slots(S_1) \times Slots(S_2)$, enabling transfer between structurally distinct representations, subject to constraint alignment.
• Path mappings: To reconcile structural mismatches in data or schema graphs, path mappings are defined as $Path_o \equiv_p Path_g$ when the start and end classes are subsumed, extended with SWRL rules for executable remapping (Berges et al., 22 Jan 2024).
• Ontology-driven alignment: Models are "unpacked" into explicit ontological commitments via ontological unpacking, mapping symbols not only by name but by underlying truthmaker (entity) semantics (Guizzardi et al., 2023).
• Information Flow (IF) quotienting: Community ontologies are fused via coproducts and equivalence quotients, yielding a merged "virtual ontology" that harmonizes types and instances (Kent, 2018).
• Graph-theoretic metrics: Centrality analysis is used to identify "pivot" attributes or models for the core of a layered mesh. Interfaces are then designed to map domain-specific extensions onto these core pivots (Staebler et al., 2023).
• Competency questions and design patterns: These guide both the extraction of consistent submodels and the reuse of modeling patterns across layers (Beverley et al., 14 Jun 2025).

These mechanisms support bidirectional mapping (lifting and projection), automated reasoning, and ensure alignment of both labels and semantics at all layers.

4. Practical Layered Architectures and Workflows

Layered ontological architectures are instantiated in varied domains:

• Astronomical Instrumentation: A strict two-layer scheme distinguishes a heavyweight OWL meta-model layer (classes, properties, rules) and a lightweight model layer (individuals, facts), delivering run-time semantic layers via OPC UA (Pessemier et al., 2013).
• Materials Science (EMMO): EMMO supplies spatiotemporal mereotopology, semiotics, and 4D realism at the top level, with marketplace/domain/sub-domain ontologies and stepwise human-guided alignments for data fusion (Horsch et al., 2020).
• Orbital Debris SSA: Modular, multi-tier ontologies: upper (philosophical), scientific reference, ODO domain, and instance data, with mediation via OWL+CLIF, mapping rules, and rigorous axiomatics (Rovetto, 2017).
• Digital Engineering (DEFII): L₀–L₄ layering (BFO–CCO–domain–application–instance) with Model Interface Specification Diagrams (MISD) as semantic interfaces for tool-agnostic operations (Dunbar et al., 2022).
• EHR Data: Canonical ontology at the apex, modular per-system ontologies, and translated repositories, bridged by mapping modules with path-mapping for structural alignment (Berges et al., 22 Jan 2024).
• Risk/Attack Trees: Foundational ontology (UFO), core domain ontology (COVER), application-level metamodels (AttackTree Ontology), enabling unambiguous mapping of modeling constructs and risk metrics (Oliveira et al., 30 Jun 2025).

Workflows integrate referential and methodological ontologies through iterative steps: capture, modeling, mapping, automated code generation, and formal verification (Beverley et al., 14 Jun 2025, Pessemier et al., 2013). Multi-layer triplestores, template engines, and DL reasoners support live, semantically consistent software and analysis artifact generation.

A general workflow emerging from network and FAIR layering involves:

1. Construction and alignment of core ontologies across domains (Staebler et al., 2023).
2. Semi-automated mapping to core pivots guided by centrality analysis.
3. Layered instantiation, with domain/app-layer models referencing only constructs defined in upper layers.
4. Exposure of run-time semantic layers for dynamic interoperability (e.g., via OPC UA, RESTful endpoints).

5. Cognitive and Human-Centric Layering

Beyond classical technical, semantic, and organizational layers, recent work introduces "cognitive interoperability"—human-friendly, explorable, layered representations (Vogt, 2023, Vogt et al., 2023). Semantic units are modeled as first-class objects (StatementUnit, CompoundUnit, QuestionUnit), creating a discursive layer decoupled from the data graph. This enables UIs and mind-maps that dynamically reflect layered knowledge and support zooming across five granularity levels—from triples to datasets (Vogt, 2023).

The FAIREr paradigm extends the FAIR principles (Findable, Accessible, Interoperable, Reusable) to include Explorability, and modular frameworks (e.g., Rosetta Stone) define a reference schemata layer as interlingua for minimization of crosswalks and maximal human intelligibility (Vogt et al., 2023, Vogt et al., 6 May 2024).

Tables of semantic units, their class, and granularity layer:

Semantic Unit Class	Description	Granularity Layer
StatementUnit	Atomic proposition	L¹ (atomic statement)
ItemUnit	All statements on a subject	L² (object-centered)
ItemGroupUnit	Group of related item units	L³ (compound structure)
DatasetUnit	Whole KG/subgraph	L⁴ (entire dataset)

6. Challenges, Open Problems, and Best Practices

Semantic interoperability via ontological layering faces persistent challenges:

• Semantic Heterogeneity: Name/structural conflicts and coverage gaps pervade domain ontologies. Central-pivot layering and use of standard core ontologies mitigate this but do not eliminate need for human oversight (Staebler et al., 2023, Rovetto, 2017).
• Expressivity Mismatches: Practical interoperability is sometimes constrained by logic expressivity limits (e.g., OWL-DL vs. full first-order or higher category-theoretic approaches), leading to pragmatic trade-offs between speed, expressivity, and maintainability (Horsch et al., 2020, Dunbar et al., 2022).
• Path/Structural Mapping Complexity: Robust path mapping (structural alignment via regular expressions, SWRL rules, etc.) is computationally nontrivial and may require heuristics, especially for complex or deeply nested schemata (Berges et al., 22 Jan 2024).
• Versioning and Curation: The maintenance, evolution, and community curation of upper and core-layer ontologies is nontrivial, with terminology drift and domain change driving continual revision cycles (Rovetto, 2017, Dunbar et al., 2022).
• Cognitive Layer Integration: Systematic, automated translation between machine-actionable layers and cognitively-accessible representations (mind-maps, templated views) remains an ongoing area of development (Vogt, 2023, Vogt et al., 2023).

Key best practices include: strict separation of concerns by layer, community-curated and openly published vocabularies, the use of pivot/core schemas for adaptation, explicit mapping and path rules, and integrated reasoning and validation pipelines (Horsch et al., 2020, Pessemier et al., 2013). For evaluation, competency questions and SPARQL queries are used to demonstrate the semantic closure and correctness of layered integrations (Rovetto, 2017, Beverley et al., 14 Jun 2025).

7. Theoretical Perspectives and Mathematical Formalizations

The abstract foundation of semantic interoperability and ontological layering can be modeled in category-theoretic terms. For example, Kent’s framework models ontologies as adjunctions and their layers as closure/interior operators, with polar factorizations yielding a “lattice of theories” for each ontology (Kent, 21 Apr 2024). Semantic integration corresponds to constructing mediating adjunctions (diamond mappings) between concept lattices of community ontologies.

Summary of categorical layering:

Layer	Categorical Structure	Role
Base/Data Layer	Set-valued	Instance data
Signature Layer	Ordered sets (A₀, A₁)	Types, properties
Conceptual Layer	Concept lattices (diamond(g))	Concept alignment
Theory Layer	LOT (Lattice of Theories)	Abstract truth structure

Such frameworks provide not only semantic clarity but theoretical guarantees of modularity, composability, and scalability (Kent, 21 Apr 2024, Kent, 2018).

---

References:

• (Pessemier et al., 2013): Pessemier et al., "A practical approach to ontology-enabled control systems for astronomical instrumentation"
• (Beverley et al., 14 Jun 2025): Tolk et al., "Ontology Enabled Hybrid Modeling and Simulation"
• (Rovetto, 2017): Allen, "An Ontological Architecture for Orbital Debris Data"
• (Dunbar et al., 2022): Dunbar et al., "Driving Digital Engineering Integration and Interoperability Through Semantic Integration of Models with Ontologies"
• (Vogt et al., 6 May 2024): Vogt, "FAIR 2.0: Extending the FAIR Guiding Principles to Address Semantic Interoperability"
• (Vogt, 2023): Vogt, "The FAIREr Guiding Principles: Organizing data and metadata into semantically meaningful types of FAIR Digital Objects to increase their human explorability and cognitive interoperability"
• (Staebler et al., 2023): Kringel et al., "Towards solving ontological dissonance using network graphs"
• (Horsch et al., 2020): Jablonka et al., "Reliable and interoperable computational molecular engineering: 2. Semantic interoperability based on the European Materials and Modelling Ontology"
• (Horsch et al., 2020): Schober et al., "Semantic interoperability based on the European Materials and Modelling Ontology and its ontological paradigm: Mereosemiotics"
• (Kent, 2018): Kent, "The Information Flow Foundation for Conceptual Knowledge Organization"
• (Guizzardi et al., 2023): Guizzardi et al., "Semantics, Ontology and Explanation"
• (Berges et al., 22 Jan 2024): Vogt et al., "Toward Semantic Interoperability of Electronic Health Records"
• (Kent, 21 Apr 2024): Kent, "The Characterization of Abstract Truth and its Factorization"
• (Oliveira et al., 30 Jun 2025): Almeida et al., "An ontological lens on attack trees: Toward adequacy and interoperability"
• (Vogt et al., 2023): Vogt, "Towards a Rosetta Stone for (meta)data: Learning from natural language to improve semantic and cognitive interoperability"