Layered Hypergraph Architecture

Documentation status: reference — see Maturity and evidence.

Conceptual memory can be read as several logical layers over the same set of concepts and relations. These layers help organization, governance, and projection without duplicating conceptual identity.

Typical layers

• vocabulary: concepts, types, and roles;
• instances: identified objects and facts;
• relations: binary or polyadic links;
• constraints/rules: validity conditions and derivations;
• contexts: scope, time, provenance, or domain;
• projections: indexes, views, persistence, or computational representations.

Identity principle

A projection does not automatically create a new business identity. Conceptual identifiers remain stable across layers when they describe the same reality.

Access

Queries should target the semantic granularity appropriate to the task and avoid depending on an index or storage detail that may change.

Evolution

New layers can be added additively when their role is explicit. Internal storage and optimization mechanisms are not public contracts.