Semantics, Platforms, and the Illusion of Control: Why Open Standards Alone Will Not Save Enterprise Meaning

Article previously published on LinkedIn

Recent discussions around platform consolidation, real-time data pipelines, and AI architecture increasingly emphasize the strategic role of formal semantics. The argument is compelling: as cloud providers, streaming platforms, and AI stacks become vertically integrated, enterprises risk losing not only control of infrastructure, but also control over the meaning of their own data. In this context, open semantic standards such as RDF, OWL, and SPARQL are often presented as the last sovereign layer that organizations can truly own.

While this diagnosis captures a real and growing risk, it also underestimates the depth, complexity, and governance challenges of the semantic problem itself. Formal semantics is necessary — but far from sufficient — to preserve enterprise independence, knowledge sovereignty, and long-term resilience.

The Structural Limits of the W3C Semantic Stack

RDF, OWL, and SPARQL provide a powerful formal framework for representing and querying symbolic assertions. Yet they come with structural limitations that are often glossed over in strategic narratives.

RDF’s triple-based model atomizes knowledge into subject–predicate–object fragments. This atomization shifts much of the real semantics into external modeling conventions, graph patterns, and governance processes. Complex constructs such as context, temporality, causality, responsibility, intention, and organizational commitments are not native to the model and must be reconstructed indirectly.

SPARQL, despite its role in the “Semantic Web” stack, remains fundamentally a syntactic graph pattern-matching language. It does not operate on meaning in a cognitive or operational sense, but on formal symbolic structures.

OWL, for its part, is intentionally limited to decidable fragments of first-order logic. This design choice is essential for automated reasoning at scale, but it also restricts what can be expressed: full causality, rich process logic, normative constraints, and many real-world organizational semantics lie outside its formal scope.

Together, these technologies enable formal representation, but they do not constitute an operational theory of enterprise knowledge or meaning in use.

Formal Semantics Is Not a Substitute for Computation or Knowledge Management

Another frequent confusion is the implicit idea that formal ontologies could replace or subsume computational systems and knowledge management practices. They cannot.

Enterprises do not operate on ontologies alone. They rely on:

• execution platforms,
• transactional systems,
• event-processing engines,
• organizational workflows,
• human decision processes,
• governance structures,
• and institutional memory.

Ontologies and knowledge graphs describe states of affairs. They do not execute business processes, enforce accountability, or manage organizational change. Without tight integration into socio-technical systems and operational governance, formal semantics risks becoming a descriptive layer disconnected from real enterprise behavior.

Upper Ontologies: Necessary but Not a Universal Solution

Upper or top-level ontologies (BFO, DOLCE, etc.) are often presented as a solution to cross-domain coherence. They are indeed important enablers for:

• conceptual alignment,
• logical consistency,
• interoperability across disciplines.

However, they do not resolve all semantic challenges. They introduce their own difficulties:

• high modeling and philosophical expertise is required,
• significant training and institutional maturity are needed,
• costly and long-term maintenance is unavoidable,
• interpretation remains context-dependent despite formal rigor.

Upper ontologies are powerful instruments, but also demanding ones. They do not eliminate semantic risk; they transform it into a high-skill, high-governance problem.

Open Standards, Proprietary Ontologies, and the Risk of Conceptual Lock-In

Today, major cloud and AI providers are actively promoting their own simplified, proprietary ontologies and schemas as built-in semantic layers for data integration, AI services, digital twins, and knowledge graphs. Even when these are wrapped in “open” APIs, they increasingly function as de facto conceptual standards.

This creates a new and subtle form of dependency: not only on data formats or platforms, but on conceptual structures themselves. Semantic lock-in can arise even when the underlying technologies claim compliance with open standards.

The history of information systems shows that formal openness does not automatically guarantee effective sovereignty.

A Historical Reminder: The ISO STEP Experience

The ambition to decouple enterprise meaning from technology is not new. ISO STEP (10303) already pursued this for product data exchange, sharing, and long-term archiving decades ago. It relied on a layered semantic and application modeling approach (AAM, ARM, AIM) and on a strong consensus process between industrial clients and solution providers.

Yet over time, STEP suffered from:

• fragmentation,
• partial capture by technology providers,
• decreasing business involvement,
• and misalignment with the actual closed platforms later imposed by IT vendors.

Only a few major industrial actors (such as in aerospace) maintained strong governance over these models. This history illustrates a fundamental lesson: semantic independence is not preserved by standards alone, but by sustained industrial and institutional control.

Concentration, State Power, and the Geopolitical Dimension of Semantics

A deeper risk now emerges beyond vendor lock-in: the concentration of semantic infrastructures in the hands of a few global technology players, increasingly under the influence or control of state power. Data platforms, AI models, cloud infrastructures, and embedded ontologies are becoming strategic assets in geopolitical competition.

Step by step, enterprises — and even states — are losing operational and conceptual independence. This process has been underway for decades, but it is now accelerating. The resulting vulnerabilities increasingly look less like “missed business opportunities” and more like systemic weaknesses and critical security threats.

In this context, it is far from certain that semantic technologies based on open standards alone will be sufficient to “save the game.” Without strong governance, diversified technological ecosystems, and genuine industrial ownership of meaning, open standards risk becoming symbolic safeguards rather than effective shields.

Conclusion: Semantics Is a Governance Problem Before It Is a Technical One

Formal semantics is indispensable. Without ontologies, no long-term interoperability, no coherent AI reasoning, and no durable knowledge infrastructure is possible. Upper ontologies, RDF, OWL, and SPARQL are essential components of modern semantic architectures.

But they are not neutral tools, and they are not sufficient guarantees of independence, resilience, or sovereignty.

Enterprise meaning is not preserved by formal models alone. It requires:

• computation,
• platforms,
• organizational governance,
• industrial alignment,
• human competencies,
• and long-term political and economic control.

The risk today is not only technological lock-in, but conceptual, institutional, and geopolitical lock-in. If semantics is treated purely as a technical layer, it may become yet another domain where enterprises believe they are independent — while in fact they are not.