HARNESS D1.4 Legal Ontologies - version 1

Translating the context-dependent expression of a law into a format that machines can interpret remains a primary challenge. This is significantly compounded by the fact that legislative bodies primarily rely on complex, multi-stacked regulatory norms to govern digital systems, and as this practice persists, the development of machine-readable implementations has become a prerequisite for automated compliance and legal interoperability. It is within this context that Deliverable D1.4 Legal Ontologies - version 1, developed under Work Package 1 (WP1) of the HARNESS Project, can serve as an intervention. This work addresses these challenges by proposing a foundational architectural structure that bridges the gap between legal requirements and their machine-readable representation. Significantly, a primary priority is of defining a modular knowledge layer that can serve as the indispensable basis for subsequent project iterations and ensures that advancements in ontological complexity are built upon a scalable conceptual architecture.

The complexity of translating a statutory norm into computational logic is primarily exacerbated by (i) the linguistic variations of a single legal concept exhibiting divergent and ambiguous legal meanings (known as Semantic Fragmentation) and (ii) the uncoordinated rules of independent legislative acts that create contradictory compliance constraints (known as Normative Collision). Given these systemic challenges, D1.4 Legal Ontologies - version 1 takes a lightweight approach to maintain coherence while allowing for incremental refinement. This informs for the development of strategies as defined within D1.4 Legal Ontologies - version 1 to ensure that the foundational work persists robust to support future scaling and adaptation to emerging legal requirements. The methodology utilises the Agile phases of the Linked Open Terms (LOT)  to define a core semantic structure using semantic standards such as RDF  and SKOS  to support the transition from natural-language legal expressions to a standardised machine-readable format grounded in best practices. The current iteration demonstrates this through the European Health Data Space (EHDS) , however, the methodology is designed for broader applicability across a wider legal corpus, specifically the General Data Protection Regulation (GDPR) , the Artificial Intelligence Act (AIA) , the Data Act (DA) , the Data Governance Act (DGA) , the Charter of Fundamental Rights across the European Union (EUCFR) , and the Network and Information Systems Directive 2 (NIS2) . Significantly, the empirical extraction, semantic mapping samples, and structural verification presented reflect the legal concepts and methodological rigor developed through a research stay at Ghent University . Moreover, this work introduces two methodological mechanisms that can assist in justifying convergence and divergence patterns existing within Obligation-Obligation pairs. These mechanisms can, therefore, accommodate for the increase of the technical sophistication of the work within following iteration advancements, without compromising semantic integrity. 

Central to the ontological foundation is the formalisation of the Actor, Right, Obligation legal concepts. Grounding the knowledge layer in these legal concepts, the architecture designs how Actors interact within a legal domain, identifying the Rights they hold and the compliance Obligations to which they are bound to fulfil. The deliverable operationalises these legal concepts to move beyond abstract conceptualisation and deliver a lightweight architecture, providing the project with transparent documentation extensible to the remainder of the EU-level legal frameworks involved in the project deliverable. This approach, ultimately, sets the foundational logic that can be utilised for future refinements towards a unified, cross-regulatory ontology, which (i) defines the core legal concepts for Actors, Rights, Obligations and (ii) semantically aligns their technical convergence and divergence across different EU regulatory norms. By leveraging established vocabularies (e.g., W3C Data Privacy Vocabulary (DPV) ) as a base for semantic modelling, the future Knowledge Graph (KG) will seek to achieve interoperability and remain responsive to evolving legal requirements.