GPT Review — Object Relationship Layer, Điều 39 KG, and IU Compatibility

Date: 2026-05-28 Reviewer: GPT Council via Web Connector fallback

Sources checked via Agent Data search

• knowledge/dev/architecture/layer3-information-law.md — each object has its own DB/layer 3 information surface.
• knowledge/dev/architecture/information-atom-law.md — governed object conditions and 8 relation rules: IDENTITY, BELONGS_TO, CONTAINS, DEPENDS_ON, USED_BY, TRANSITIVE, PEERS, SIMILAR.
• knowledge/dev/architecture/species-taxonomy-complete.md — species taxonomy + birth registry + collection registry pattern.
• knowledge/dev/laws/dieu39-knowledge-graph-law.md — enacted KG law; 26 problems, 36 DOT/18 pairs, trusted KG for AI decisions.
• knowledge/dev/laws/dieu44-trien-khai/design/04-information-unit-profile-schema.md — IU as universal substrate / living DB / UMC / DOT contract.
• knowledge/dev/laws/dieu44-trien-khai/design/07-iu0-information-unit-minimum-standard-outline.md — IU minimum standard.
• knowledge/dev/laws/dieu38-trien-khai/tham-khao/02c1-text-unit-catalog.md — vertical structural relations vs horizontal reference graph distinction.
• knowledge/dev/reports/architecture/iu-b-c-f-pure-readonly-live-proof-2026-05-28.md — b/c/f live proof, axis/tree/sql links.
• knowledge/dev/reports/architecture/iu-b-c-f-additive-dot-wrapper-and-harness-live-apply-2026-05-28.md — b/c/f DOT-callable live proof.
• knowledge/dev/reports/architecture/iu-mutation-safety-foundation-for-d-e-f-enable-live-apply-2026-05-28.md — gate protocol and readiness for mutation tests.

Verdict

DESIGN_COMPATIBLE_BUT_KG_RELATIONSHIP_LAYER_NOT_YET_FULLY_LIVE_FOR_IU

The object relationship doctrine exists and is compatible with the current IU direction. It has been designed at the architecture/law level: every governed object eventually has a DB/information surface about itself and standardized relationships such as belongs_to, contains, depends_on, used_by, peers, similar.

For Information Units, the currently proven live substrate covers important relationship subsets:

• tree/parent-child/ancestor relation via IU tree/subtree structures;
• professional/domain tags via axis-B metadata;
• SQL link relations to live PG objects;
• collection/membership relations;
• structure operation readiness;
• event/gate readiness for future trigger tests.

However, the full Điều 39 KG relationship layer is not yet fully live for IU. The current IU relationship layer is sufficient for tests b/c/f and ready to proceed to d/e/f-enable, but the richer KG relations — who uses me, whom I use, ownership/usage/dependency/provenance/negative/temporal enrichment — should be added as a future KG/IU enrichment phase after the IU tests close.

What is already there conceptually

1. Object-as-own-DB / Layer 3 doctrine: each object has a DB/information surface about itself.
2. Governed object law: identity, registry, metadata, lifecycle, owner, relationships.
3. 8 base relationship rules: IDENTITY, BELONGS_TO, CONTAINS, DEPENDS_ON, USED_BY, TRANSITIVE, PEERS, SIMILAR.
4. Species taxonomy / birth registry: objects are classified from birth, avoiding wrong cage/orphan entities.
5. Điều 39 KG: future trusted graph for AI decision-making, with DOT/governance.

What is live/proven for IU

1. Axis-B professional/domain filter PASS and DOT-callable.
2. Axis-C parent-child-grandchild tree PASS and DOT-callable.
3. SQL links validate/resolve PASS and DOT-callable.
4. Gate protocol is live, enabling future governed mutation tests.
5. IU substrate already has enough relationship structure to continue d/e/f-enable tests.

What is not yet fully live

1. A unified generic KG edge table for all IU/object relations is not proven live in this review.
2. Standard inverse relationships such as uses/used_by, belongs_to/contains, depends_on/dependency_of are conceptually defined but not proven as a complete live IU enrichment mechanism.
3. Automatic relationship enrichment from events, task outcomes, Kaizen suggestions, SQL links, and workflow usage is not yet proven.
4. Temporal/provenance/negative knowledge from Điều 39 is not yet integrated into IU test flow.
5. IU process-brick fields such as kg_edge_refs remain design/readiness targets, not proven complete live substrate.

Recommendation

Do not block the current IU d/e/f-enable test path on full KG implementation.

Continue current sequence:

1. Finish IU d/e/f-enable governed tests.
2. Close IU production-readiness evidence.
3. Then run a dedicated IU_KG_RELATIONSHIP_ENRICHMENT_READINESS_AND_DESIGN_PATCH_*X macro to map Điều 39 relationship primitives into IU/object relationships.

The future KG/IU macro should define:

• standard relation vocabulary for IU and all object types;
• inverse pairs and derivation rules;
• edge provenance and confidence;
• temporal validity;
• event/task/SQL/Kaizen sources that enrich relationships;
• PG-first storage and DOT commands;
• how relations project to Điều 39 KG without duplicating canonical object state;
• how 4 Mothers use those relations.

Strategic conclusion

The user's intuition is correct: since every object has its own information surface and relationships are born with the object, Điều 39 improvements must eventually enrich those object relationship layers, including IU. But for the immediate IU test roadmap, the existing relationship subset is adequate: b/c/f have passed, and d/e/f-enable should proceed. Full KG enrichment is a next phase, not a prerequisite for the remaining IU tests.