The Deterministic Routing Fabric 

for Addressable Ontology Networks 

IRISweb is a system designed to store and organize ontologies in a fully addressable and navigable structure. It is not a database, a compute engine, or a knowledge model in the traditional sense. Instead, IRISweb provides a framework in which ontologies can reside, be discovered, and be read in a controlled, deterministic manner, without requiring computation or execution of logic. The network allows ontologies to exist on local storage, private networks, distributed systems, or public platforms such as GitHub, and makes them fully resolvable through a standardized pointer system.

Every IRISweb installation begins with a small, minimal YAML file called the Root Catalog. This catalog contains no logic, no data, and no executable code. Its purpose is to act as a directory of pointers to ontology files. The Root Catalog is the entry point for the network, and its minimalism ensures that systems can start from nothing while remaining fully navigable. From this point, the network grows recursively, resolving sub-manifests, ontology shards, and structural elements only as they are needed.

This design prevents the preloading of unnecessary context or the accumulation of redundant information, allowing the network to remain light and precise regardless of scale.

IRISweb is structured as a recursive network. Each pointer in the Root Catalog resolves to a sub-manifest, which in turn can contain additional pointers, forming a fractal-like hierarchy of ontology shards. This approach allows the system to represent arbitrarily complex ontologies without requiring the full structure to be loaded at any given time. Traversal is driven by intent, which is mapped through the network by a resolution lens. Only the structures required to satisfy that intent are read into context. Once the relevant paths are traversed, the structures can be released, leaving no residual state in the system.

The network is organized through clearly defined topological units. A Node represents the atomic unit of the network and is assigned a unique identity via SHA256 hashing. Nodes are grouped into Clusters, which are collections of related nodes; clusters may be structural or governed by a meta-ontology. Hubs organize clusters, providing an indexable structure that allows content to be located efficiently. Routers follow pointers through nodes, clusters, and hubs, resolving paths recursively. Matricesrepresent networks of routers, forming the routing fabric of IRISweb, while Gateways provide the initial resolvable entry points into the network.

Each of these units is fully addressable and externally identifiable, ensuring that the network is transparent, verifiable, and reconstructable at any point in time.

IRISweb does not store session memory. Instead, it provides a controlled method of loading ontology structures. The system “hydrates” only the shards required for a particular purpose. These shards are read from their external storage locations, assembled as needed, and made available for inspection or traversal. After use, the loaded structures are released, restoring a clean state. This method allows IRISweb to provide the advantages of persistent knowledge without introducing the risks of context drift or accumulation of unnecessary data.

Because IRISweb relies entirely on externally stored, addressable structures, it is inherently stateless. No session information persists within the network itself. Yet, it is deterministic: the same pointer path will always resolve to the same structures, allowing ontologies to be revisited, audited, and verified consistently. This makes IRISweb suitable for environments where reproducibility, structural integrity, and transparency are essential.

IRISweb provides the substrate for ontologies that define behavior or structure. For example, XERO ontologies can be stored and read within IRISweb. The network ensures that ontologies are accessible, composable, and navigable. Authoring tools such as XERODoc can create, update, and version ontologies within the IRISweb framework, providing a bridge between human intent and machine-readable structure. The network does not interpret or execute ontologies; it ensures that they are addressable and consistently retrievable.

• The network is a map, not a book. It provides structural navigation rather than a linear document.  
   
• Logic and behavior are defined externally; IRISweb only stores and exposes structure.  
   
• Complexity is managed through depth and recursion rather than preloading or accumulation.  
   
• Poor outcomes are not the result of the network, which guarantees integrity, but of the quality of authored ontologies.  
   

IRISweb allows systems to operate without persistent session state while still supporting large, complex, and distributed ontologies. By separating storage, addressing, and traversal from execution, it enables precise, reproducible, and scalable knowledge representation. Systems can be expanded incrementally, and users can inspect or navigate ontologies without risk of context contamination or unintended side effects.

IRISweb is designed to be location-independent. It can function on a single local machine, across private networks, or in distributed systems spanning multiple servers or cloud platforms. Its only requirement is that pointers resolve to storage locations accessible to the system. This flexibility allows IRISweb to support diverse use cases, from local ontology development to globally distributed knowledge networks.

IRISweb is a network for structured ontologies. It provides a rigorous, deterministic, and fully addressable environment in which ontologies can reside and be read. By emphasizing structure, integrity, and navigability, IRISweb enables systems to manage complex knowledge artifacts efficiently, reliably, and without relying on session state or execution logic.