KB-6B7C

KG/dot-kg Discovery — 02 KG Content Cluster Map

7 min read Revision 1
kgcontent-mapworkstream-a2026-06-04

02 — KG Content Cluster Map (Workstream A)

Source for every IU: knowledge/dev/laws/dieu39-knowledge-graph-law.md (Điều 39 — Luật Knowledge Graph v2.3, BAN HÀNH). No IU body edited; mapping only. All 16 IUs read live (information_unit, doc_code='knowledge/dev/laws/dieu39-knowledge-graph-law.md', unit_kind=law_unit, lifecycle_status=draft).

2.1 All 16 KG IUs in document order

sort IU id section section_type subtheme content class
1 f1273184 #title heading meta
2 3676cc0c #0-tam-nhin section Vision (5-tier value chain) law/canon
3 09682146 #1-26-bai-toan section 26 problems (A/B/C/D groups) law/canon
4 a448e397 #2-dinh-nghia section Definitions (KG, triple, ontology, trust) law/canon
5 620f7646 #3-4-chi-tiet-nhom-a-b section Detail groups A (infra) + B (business) design
6 e97fc334 #5-chi-tiet-nhom-c section Detail group C (architecture C1–C13) design
7 d933ec87 #6-10-quy-trinh process 10 processes A–J (the workflow spec) law/process
8 10649925 #7-36-dot-kg technical_spec 36 DOTs / 18 pairs (the bridge) law/process
9 e5d9bed3 #7b-bien-phap-quan-ly-tham-quyen governance_process Governance & authority (Đ37) governance
10 1b0f91d3 #7c-moi-truong-thuc-thi technical_spec Execution environment (PG config tables) ops/technical
11 a7bba115 #8-kien-truc technical_spec Architecture (stack, 5-tier sync) ops/technical
12 57553011 #9-11-nt principle 11 principles (NT) law/canon
13 f17d8fda #10-roadmap process Roadmap (phase 0–4) design
14 c96b53fd #12-quan-he reference_mapping Relations to other Điều reference
15 a8e7f4db #13-no-ky-thuat open_decision_list Technical debt (39 items) design/open
16 1cded144 #changelog changelog Changelog (v1.0→v2.3) reference

2.2 Subtheme grouping (the cluster's internal shape)

  1. Foundations (why + what): #0 vision, #1 26-problems, #2 definitions, #9-11 principles. → the canon layer.
  2. Design detail (how, conceptual): #3-4 groups A/B, #5 group C (C1–C13), #10 roadmap. → the design layer.
  3. Process & automation (how, operational): #6-10 the 10 processes A–J, #7-36-dot-kg the 36-DOT matrix. → the process layer — this is the dot-kg bridge.
  4. Governance & environment: #7b authority (Đ37), #7c execution env (PG config tables), #8 architecture. → the operating layer.
  5. Cross-reference & history: #12 relations, #13 tech-debt, #changelog. → the reference layer.

2.3 Sequence / logic (from the law)

The law's own logic is a continuous bottom-up ↔ top-down loop (§6.0 "Bức tranh lớn"): Data(PG) → Data Graph → Knowledge Graph → Priority Graph → AI decision → Human oversight, with the 10 processes A–J realising each arrow (A formation, B business, C feedback, D monitoring, E scaffold, F priority, G discovery, H eviction, I self-learning, J conversational). The golden rule governs all of it: "AI ĐƯỢC ĐỀ XUẤT, KHÔNG ĐƯỢC TỰ BAN HÀNH TRI THỨC CHUẨN" — TBox (schema/ontology) is human-approved; ABox (data/link/weight) is AI-automated, with the auto/manual boundary in PG kg_auto_approve_rules.

2.4 Content classification

  • Law / canon (must not be paraphrased as final unless cited): #0, #1, #2, #9-11, #6-10, #7-36-dot-kg. These are BAN HÀNH (enacted) — but every IU's lifecycle_status is still draft in information_unit. Drift flag: the document header says BAN HÀNH (v2.3, Council 3-round approved) while the IU rows are draft. Treat as enacted-in-text, draft-in-registry.
  • Design notes: #3-4, #5, #10-roadmap, #13-no-ky-thuat.
  • Operational notes: #7c execution env, #8 architecture.
  • Governance: #7b authority.
  • Reference: #12, #changelog.

2.5 Relation to DOT / process

  • #7-36-dot-kg (IU 10649925) ↔ the 36 live DOT_KG_* rows — direct, the bridge. The matrix in this IU = the 18-pair structure in dot_tools.paired_dot.
  • #6-10-quy-trinh (IU d933ec87) ↔ the 10 processes A–J — the workflow spec the DOTs implement. Also the IU shared with the workflow and cut_pipeline topic candidates.
  • #7c execution env (IU 1b0f91d3) ↔ the 14 PG config tables (kg_thresholds, kg_signal_config, kg_auto_approve_rules, …) the DOTs read each run.

2.6 Missing parts (content gaps)

  1. No per-DOT spec IUs. The 36 DOTs are described only in one matrix IU; there is no IU (or dot_tools.description/script_path) detailing each DOT's exact input/output/SQL. D2 must reconstruct from the law's process A–J text.
  2. No KG config-table reference IU. The 14 PG config tables are named in #7c but not individually documented.
  3. No runtime/operations IU. Nothing documents how the pg_function execution actually fires (it never has — 0 executions).
  4. No bridge to knowledge_documents. The KG law has no derived knowledge_documents yet (the topic→document bridge is unpopulated).

2.7 Contradictions / drift

  • Count drift: "18 DOTs" (reports/memory) vs 36 (live + law). Resolved → 36.
  • Naming drift: law uses flat DOT-KG-{ACTION} + processes A–J; live DB adds dotted kg.formation-style domain. The dotted domains are an overlay, not law.
  • Lifecycle drift: header BAN HÀNH vs IU rows draft.
  • Execution drift: execution_engine='pg_function' declared, but script_path/file_path NULL and 0 executions — declared engine with no realised execution.

2.8 D1 outline (drives doc 03)

  1. Title, purpose, scope, candidate-status warning.
  2. Source IUs (the 16) + lifecycle/drift note.
  3. KG vision & value chain (#0) + the 26 problems (#1).
  4. Definitions & trust model (#2) — survival gate + trust_score formula.
  5. Architecture C1–C13 summary (#5) + execution environment (#7c, the 14 config tables) + stack/sync (#8).
  6. The 10 processes A–J (#6-10) → relation to the 36 dot-kg DOTs (#7-36) → cross-ref to D2.
  7. Governance & authority (#7b, Đ37) + the golden rule (TBox/ABox boundary).
  8. Relation to Registries-Pivot (topic axis AX-TOPIC, process axis AX-PROCESS, the joint map).
  9. Current gaps (§2.6) + drift (§2.7) + next actions.
  10. Evidence references.
Back to Knowledge Hub knowledge/dev/reports/architecture/kg-dot-process-discovery-and-document-building-pilot-2026-06-04/02-kg-content-cluster-map.md