07b44b0920
Brief 1 scopes the PR #97 follow-up: flip `definitional_layer` to false on the three new domain seed packs (en_mathematics_logic_v1 / en_physics_v1 / en_systems_software_v1) so the ADR-0084 closure verifier stops flagging them as malformed. They're governed by ADR-0091 `domain_contract_version`, which is a different layer; the integration test's allowlist already excludes them, so the flag-flip aligns the script with the test's clearly-intended scope. Brief 2 scopes ADR-0073 L1.1 content phase II: extend the grc/he anchor-lens substrate with 8 cross-language lemma pairs (νοῦς ↔ בינה / διάνοια / καρδία ↔ לב / ψυχή ↔ נפש / ἔλεος / εἰρήνη / δικαιοσύνη / ἅγιος ↔ קדוש) authored against the L1.1 file shapes, alignment edge weights, and en-collapse annotation discipline. Anchor-lens-tour seam claims and cognition-eval byte-identity under unanchored default remain the gates. Both briefs are bounded to content authoring; neither touches code.
125 lines
No EOL
5.2 KiB
Markdown
125 lines
No EOL
5.2 KiB
Markdown
# Codex Brief - Greek/Hebrew Content Phase II
|
|
|
|
**Audience:** Same agent or content partner that seeded the original grc/he cognition packs.
|
|
|
|
**Mission:** Deepen the cognition-tier Greek and Hebrew packs by adding another layer of tightly scoped, semantically aligned lemmas. The current seam is ready; the packs are still skeletal. The target is to move each pack from the current small seed into a more usable content tier without changing the architecture or the runtime.
|
|
|
|
This is a content-only brief. Do not add new infrastructure, new loaders, or new runtime behavior.
|
|
|
|
**Target packs:**
|
|
|
|
- `language_packs/data/grc_logos_cognition_v1`
|
|
- `language_packs/data/he_core_cognition_v1`
|
|
|
|
**Estimated effort:** One content pass per pack plus validation.
|
|
|
|
---
|
|
|
|
## Background
|
|
|
|
The current grc/he cognition packs were intentionally seeded small. They established the core semantic seam and the cross-language collapse lines, but they are still underpopulated relative to their intended teaching and grounding role.
|
|
|
|
The useful next step is not architecture. It is content depth: more lemmas, better semantic clustering, and cleaner cross-language alignments to existing English cognition lemmas.
|
|
|
|
The working target is to deepen each pack by about 20-30 lemmas, organized by semantic family, while keeping the existing contract shape intact.
|
|
|
|
## Phase 0 - Baseline
|
|
|
|
1. Count current lemmas in both packs.
|
|
2. Inventory existing semantic families and alignments.
|
|
3. Confirm the current cross-language anchors that already exist in the pack alignments.
|
|
4. Capture the baseline counts before editing.
|
|
|
|
Suggested baseline commands:
|
|
|
|
```bash
|
|
python3 - <<'PY'
|
|
from pathlib import Path
|
|
import json
|
|
|
|
for pack in [Path('language_packs/data/grc_logos_cognition_v1'), Path('language_packs/data/he_core_cognition_v1')]:
|
|
lexicon = pack / 'lexicon.jsonl'
|
|
count = sum(1 for line in lexicon.read_text(encoding='utf-8').splitlines() if line.strip())
|
|
print(pack.name, count)
|
|
PY
|
|
```
|
|
|
|
## Phase 1 - Greek expansion
|
|
|
|
Add 20-30 new grc lemmas that cluster around the semantic families already implied by the existing seed:
|
|
|
|
- intellect and cognition: `νοῦς`, `λόγος`, `σοφία`, `γνῶσις`, `σύνεσις`, `ἐπιστήμη`
|
|
- spirit and breath: `πνεῦμα`, `ψυχή`, `πνοή`
|
|
- truth and clarity: `ἀλήθεια`, `σαφήνεια`, `φανερότης`
|
|
- ordering and measure: `τάξις`, `μέτρον`, `κρίσις`
|
|
- relation and response: `ἀπόκρισις`, `κοινωνία`, `πίστις`
|
|
|
|
For each new lemma:
|
|
|
|
1. Keep the surface and lemma stable and explicit.
|
|
2. Assign semantic domains that fit the existing pack vocabulary.
|
|
3. Add or update the relevant cross-language alignment to an existing English cognition lemma when there is a stable target.
|
|
4. Avoid forcing an English collapse when the concept does not genuinely collapse cleanly.
|
|
|
|
Example alignment targets can reuse existing English cognition anchors, such as conceptual entries around mind, word, wisdom, knowledge, spirit, truth, and judgment.
|
|
|
|
## Phase 2 - Hebrew expansion
|
|
|
|
Add 20-30 new he lemmas that cluster around the semantic families already implied by the existing seed:
|
|
|
|
- spirit and breath: `רוח`, `נשמה`, `נפש`
|
|
- heart and will: `לב`, `רצון`, `כוונה`
|
|
- wisdom and discernment: `חכמה`, `בינה`, `דעת`
|
|
- truth and faithfulness: `אמת`, `אמונה`, `נאמנות`
|
|
- covenant and steadfastness: `חסד`, `ברית`, `צדק`
|
|
|
|
For each new lemma:
|
|
|
|
1. Keep the morphology and lemma choices conservative.
|
|
2. Use semantic domains that match the current pack design.
|
|
3. Align to the closest existing English cognition lemma only when the mapping is stable and honest.
|
|
4. Preserve the existing split between direct collapse and non-collapse relation types.
|
|
|
|
## Phase 3 - Cross-language alignment
|
|
|
|
For each newly added lemma, add the most appropriate alignment annotation to the existing English cognition layer.
|
|
|
|
Rules:
|
|
|
|
1. Prefer one clear alignment over several weak ones.
|
|
2. Use non-collapse relations when English does not genuinely preserve the distinction.
|
|
3. Reuse the current evidence style already present in the pack.
|
|
4. Do not widen the English pack just to make an alignment easier.
|
|
|
|
## Phase 4 - Validation
|
|
|
|
After the content pass:
|
|
|
|
1. Recount both packs and confirm the intended growth.
|
|
2. Verify the lexicon and alignment files are internally consistent.
|
|
3. Run the pack validation and relevant cognition checks.
|
|
4. Confirm the existing runtime/eval lanes are unchanged by the content-only update.
|
|
|
|
Suggested checks:
|
|
|
|
```bash
|
|
python3 -m pytest tests/test_pack_glosses_content.py -q
|
|
python3 -m pytest tests/test_domain_pack_contract.py -q
|
|
python3 -m core.cli eval cognition
|
|
```
|
|
|
|
If any update forces an architectural change, stop and split that out. This brief is for content deepening only.
|
|
|
|
## Deliverables
|
|
|
|
1. Expanded `grc_logos_cognition_v1` lexicon/alignment content.
|
|
2. Expanded `he_core_cognition_v1` lexicon/alignment content.
|
|
3. Validation output showing the packs still load and the cognition lane is stable.
|
|
|
|
## Hard rules
|
|
|
|
- Do not change runtime code.
|
|
- Do not add new loaders or new pack kinds.
|
|
- Do not change the existing cross-language contract shape.
|
|
- Do not let English collapse convenience override semantic honesty.
|
|
- Do not turn the update into an architecture project. |