Theory of Semantic Computing Infrastructure

Modern computing systems exhibit a structural discontinuity between two domains: the execution substrate of operating systems and the reasoning substrate of artificial intelligence. Operating systems manage computational resources such as memory, processes, and storage, while AI systems operate primarily on unstructured linguistic or symbolic representations of meaning. Between these domains lies an unformalized layer: the infrastructure required to represent, govern, and compute over meaning itself. This paper introduces the concept of Semantic Computing Infrastructure (SCI) as the missing architectural layer between physical computation and cognitive systems. SCI provides mechanisms for representing meaning as executable structures, enforcing semantic governance, canonicalizing conceptual identity, and preserving lineage across semantic transformations. The paper proposes that the absence of shared governed semantic infrastructure can contribute to unsupported generation, reasoning drift, provenance gaps, and fragile integration, while recognizing that these failures have additional model, data, interface, evaluation, and institutional causes. By introducing structured services for semantic representation, identity, governance, and transformation, SCI aims to make identified knowledge representations a more explicit operational domain. The thesis is architectural: modern computing lacks a stable layer for meaning. Semantic Computing Infrastructure proposes a way to address this gap, allowing cognitive systems to query and transform governed knowledge representations while remaining bounded by their evidence and authority.