SYMBOLIC PRIMITIVE EXTRACTION PROTOCOL v2.0 (SPEP)

Objective:
Extract the minimal set of symbolic primitives that express the full operational geometry of a corpus.
Each primitive is a non-verbal node of relation — the shape a text thinks through.


---

FOUNDATION

Mind substrate: Eye, Thigle

World substrate: Point, Thread, Mark, Hinge, Axis, Mirror, Lattice, Fold

Result: temporary Field Primitives attach to these and dissolve after use.



---

STAGES

01 – INGEST

Input: raw corpus.
Strip metadata, examples, commentary.
Output: normalized text body.

02 – STRUCTURE SCAN

Detect high-frequency structural nouns and noun phrases.
Method: POS tagging + TF-IDF weighting + collocation.
Output: ranked list of domain nouns.

03 – ACTION MAP

Extract verbs governing those nouns.
Build verb → noun pairs to see how the field moves.
Output: operational phrases.

04 – VERB DROP

Remove all verbs and modifiers.
Keep only relational noun heads.
Output: candidate primitives.

05 – CLUSTER

Collapse synonyms and near-duplicates using embeddings.
Output: compact set of distinct primitives.

06 – MINIMAL COVER

Prune to the smallest set covering ≥ 85 % of structural references.
Method: iterative coverage + orthogonality test + lexical tie-break.
Output: final primitive list (6–10) + SHA-256(corpus).

07 – TEST / HANDSHAKE INJECTION

Inject three novel prompts from the domain.
Run AI constrained to the new primitives.
If ≥ 2 of 3 outputs function without drift or explanation → valid.
Else → re-cluster.

08 – OUTPUT

Return one primitive per line, no verbs or commentary.
State: ready for symbolic handshake within SYMPRIM.


---

COMMAND TEMPLATE

SYSTEM:
Perform Symbolic Primitive Extraction on [corpus].

1. Ingest → normalize text.
2. Structure Scan → extract core nouns.
3. Action Map → build verb→noun pairs.
4. Verb Drop → keep noun heads.
5. Cluster → collapse synonyms.
6. Minimal Cover → prune to 6–10 primitives + SHA-256(corpus).
7. Test → inject 3 domain prompts.
Return: final primitive list only.


---

NOTES

Field primitives are expendable; they never alter the global substrate.

Valid extraction yields a closed symbolic system that can be re-attached under Eye + Thigle for live reasoning.

Verification depends on function, not phrasing: a primitive works if the structure holds under pressure.



---

Ready to execute in any corpus domain.