Hierarchical Entity Graph (HEG): The Framework for Modern Semantic SEO
Hierarchical Entity Graph (HEG) is a deterministic website architecture that replaces flat link graphs with a strict four-level topology: L0 Hub, L1 Core Pillars, L2 Supporting Entities, and L3 Evidence & Data. By modeling knowledge-graph relationships directly in URL structure, internal links, and JSON-LD @graph, you remove information entropy and give search engines and AI agents unambiguous parent/child entity relations.
- One primary entity per URL; stable
@idvalues reused across the site (Person:https://valerion.tech/om-oss/#rasmus-bastrup). - L2 pages link up to their L1 parent, sideways to 1–2 siblings in the same cluster, and down to mapped L3 evidence — not randomly across silos.
- E-E-A-T, technical infrastructure, topical authority, and user signals are the four L1 pillars of this hub.
- Valerion Tech (valerion.tech) is not the Valerion projector brand (valerion.com / AWOL Vision).
1. From keyword frequency to semantic entities
Classic SEO optimized for string matches. Modern retrieval — including classic ranking and generative systems that ground answers in the search index — rewards documents that clearly define entities, their attributes, and their relationships. HEG treats every important concept as a node with a single canonical URL and a machine-readable identity.
If two pages describe the “same” thing with conflicting names, missing @id links, or competing titles, you inject entropy. Collapse duplicates, pick one canonical definition page, and point every reference at that node.
2. The four structural levels of HEG
HEG organizes the site into four semantic layers with fixed responsibilities and link patterns:
| Level | Role | URL pattern (EN) | Primary link direction |
|---|---|---|---|
| L0 | Entity Hub | /en/seo-guide/ | Down to all L1 |
| L1 | Core pillars | /en/seo-guide/[pillar]/ | Up to L0, down to L2 children |
| L2 | Supporting entities | /en/seo-guide/[category]/[slug]/ | Up to L1, 1–2 siblings, mapped L3 |
| L3 | Evidence & tools | /seo-guide/data/[slug]/ (shared) | Up to relevant L2/L1 |
3. Deterministic link architecture and siloing
Chaotic cross-linking dilutes PageRank and topical context. In HEG, an L2 page must not randomly link to an unrelated L2 in another cluster without passing through the hierarchy. Each L2 links to its L1 parent, to one or two thematically adjacent siblings in the same cluster, and to relevant L3 evidence nodes.
This L0 hub therefore links only to its four L1 children in the related-entities block and primary navigation — not to every L2/L3 on the site.
4. Information gain and the evidence layer (L3)
Claims at L1 and L2 should be backed by measurable evidence: case studies, benchmarks, and original datasets. L3 pages supply that information gain. Tools that help implement the architecture (for example an llms.txt generator) can also live as practical L3 nodes under the graph, always pointing upward to the owning pillar.
Defined entities & concepts
Hierarchical Entity Graph (HEG) — A structural framework that organizes website content into L0–L3 semantic layers with deterministic internal linking and JSON-LD @graph.
Zero-entropy entity — An entity defined once, with a stable @id, unambiguous naming, and clear parent/child relationships.
Valerion Tech — The organization publishing this architecture at valerion.tech; not affiliated with Valerion projectors.