CRIMSON HEXAGONAL ARCHIVE: TERM INDEX WORK PLAN

EA-REGISTRY-TERMINDEX-PLAN v1.0

Author: Lee Sharks (ORCID 0009-0000-1599-0703)

Date: 16 June 2026

Status: Work plan with progress tracking

Purpose: Systematic extraction, canonicalization, and versioning of all coined terms, concepts, entities, frameworks, operators, institutions, heteronyms, and designations across the Crimson Hexagonal Archive (~841+ deposits)

Rationale

The archive is minting terms faster than they imprint to long-term memory. An estimated 200-300 coined concepts exist across 841+ deposits without a unified index. Once built, the index becomes:

• A retrieval surface (each term becomes a searchable node)
• A density map (which terms cluster, which are orphaned)
• A versioning base (new deposits append; the initial build is the hard part)
• An MPAI generator (each indexed term can receive a provenance coupling kernel)
• A measure of the archive's actual vocabulary (what it has coined vs. what it thinks it has coined)

---

Phase 1: Metadata Pull

Status: NOT STARTED

Estimated compute: 30-45 minutes

Resumable: Yes — paginate via Zenodo API, save after each page

1.1 Pull all records from crimsonhexagonal community

• API endpoint: https://zenodo.org/api/records?communities=crimsonhexagonal&size=200&page=N
• Expected: ~841 records across 5 pages
• Save: JSON file with record ID, DOI, title, description, keywords, creators, publication_date, version, related_identifiers
• Output: termindex-metadata-raw.json

1.2 Extract terms from metadata fields

• Parse each record's title, keywords, and description
• Extract: capitalized multi-word phrases, EA-* codes, quoted terms, terms in bold/strong tags
• Output: termindex-metadata-terms.json — rough list with source DOI for each term

1.3 Deduplicate and categorize

• Categories: metric, operator, framework, institution, heteronym, journal, concept, designation, entity, tool
• Canonical form + variants
• Output: termindex-metadata-canonical.json

Progress checkpoint: After Phase 1, we have ~60-70% of coinages from metadata alone. Save all three files to /home/claude/ and present. If session compacts here, next session loads these files and proceeds to Phase 2.

---

Phase 2: File Content Extraction

Status: NOT STARTED

Estimated compute: 2-3 hours (may require multiple sessions)

Resumable: Yes — track which record IDs have been processed

2.1 Download and read markdown files

• For each record, download .md files (skip PDFs, images, audio)
• Read each file, extract terms not already in Phase 1 output
• Track progress: termindex-file-progress.json — list of processed record IDs
• Output: termindex-file-terms.json

2.2 Batch processing strategy

• Process in batches of 50 records
• After each batch: save progress, save extracted terms
• If session compacts: next session reads progress file, resumes from last batch

2.3 Merge with metadata terms

• Combine Phase 1 and Phase 2 extractions
• Identify terms that appear only in file contents (not metadata)
• Output: termindex-merged.json

Progress checkpoint: After Phase 2, we have ~90% of coinages. The remaining 10% are implicit terms that require human judgment.

---

Phase 3: Human-in-the-Loop Pass

Status: NOT STARTED

Estimated compute: 1-2 sessions of review

3.1 Lee reviews the merged index

• Flag false positives (ordinary words extracted as coinages)
• Flag missing terms (coinages Lee knows exist but weren't extracted)
• Assign canonical definitions (20 words max per term)
• Confirm categories

3.2 Cross-reference against known lists

• Dodecad heteronyms (12 positions)
• Framework numbering (FW1-FW15)
• Journal names (6 journals, 2 imprints)
• EA-* designation system
• Hex prefix allocations
• Room system (r.01-r.26)
• Operator registry (LOS, THUMB, UKTP, etc.)

3.3 Output: termindex-v1.0.json

• Full canonical index with:

term, abbreviation, category, canonical_definition

• first_deposit (DOI), author_position (heteronym)
• related_terms, hex_prefix, framework_number
• variants (alternate forms that should resolve to this entry)

---

Phase 4: Deposit and Surface

Status: NOT STARTED

4.1 Deposit to Zenodo

• EA-REGISTRY-TERMINDEX-01 v1.0
• JSON + markdown rendering
• Community: crimsonhexagonal
• Keywords: term index, coined concepts, semantic physics, operative semiotics, Crimson Hexagonal Archive

4.2 Surface deployment

• Page on leesharks.com or crimsonhexagonal.org
• Searchable, sortable table
• Each term links to its first-deposit DOI

4.3 Versioning protocol

• Each new deposit session: extract terms from new deposits, append to index
• Version forward: v1.1, v1.2, etc.
• The initial build (Phases 1-3) is the hard part; versioning is incremental

---

Phase 5: MPAI Generation (optional, high-value)

Status: NOT STARTED

5.1 For each high-priority term, generate an MPAI kernel

• Provenance coupling statement
• Non-erasure condition
• Canonical definition
• First-deposit DOI

5.2 Batch deposit MPAIs

• 10-20 per session
• Priority: terms most likely to enter the retrieval basin

---

Progress Tracking

Phase

Step

Status

Output File

Records Processed

Notes

1

1.1 Metadata pull

COMPLETE

termindex-metadata-raw.json

845/845

845 records, 6,256 unique kw

1

1.2 Term extraction

COMPLETE

termindex-metadata-terms.json

845/845

1,524 terms (count>=2), 5,951 total

1

1.3 Canonicalization

COMPLETE

termindex-tiered.json

845/845

178 Tier 1, 332 Tier 2, cross-ref with registry

2

2.1 File download

COMPLETE

termindex-file-progress.json

735/800

444 new Tier 1, 540 new Tier 2 from file contents

2

2.2 Batch processing

COMPLETE

termindex-file-progress.json

735/800

65 records had download failures

2

2.3 Merge

PENDING

termindex-merged.json

—

Needs noise filtering + human review

3

3.1 Human review

PENDING

—

—

~2,000 terms for review

3

3.2 Cross-reference

COMPLETE

termindex-crossref.json

—

129/131 registry queries matched

3

3.3 Final index

PENDING

termindex-v1.0.json

—

After human review

4

4.1 Deposit

IN PROGRESS

—

—

Initial deposit with raw data

4

4.2 Surface

NOT STARTED

—

—

5

5.1 MPAI generation

NOT STARTED

—

—

---

Continuity Protocol

If a session hits compaction limits during this work:

• All progress files are saved to /home/claude/ and presented via present_files
• The progress tracking table above is updated in this document
• The next session receives this work plan document and any saved progress files
• The next session reads the progress table, loads saved files, and resumes from the last completed step

Key files for continuity:

• This work plan: termindex-workplan.md
• Raw metadata: termindex-metadata-raw.json
• Extracted terms: termindex-metadata-terms.json
• Canonical index: termindex-metadata-canonical.json
• File processing progress: termindex-file-progress.json
• File-extracted terms: termindex-file-terms.json
• Merged index: termindex-merged.json
• Final index: termindex-v1.0.json

---

Session Log

Session 1 (16 June 2026): Work plan created. Phase 1.1 complete (845 records pulled). Phase 1.2 complete (1,524 terms extracted with count>=2; 5,951 total unique keywords). Phase 1.3 (canonicalization and categorization) ready for next session or human review. Key finding: the archive has 6,256 keyword instances across 845 records, with the top terms being Crimson Hexagonal Archive (439), semantic economy (267), Crimson hexagon (248), distributed epic (154), NH-OS (149), operative semiotics (124), training layer literature (121). The API paginates at max size=25, requiring 34 pages. The metadata-raw and metadata-terms JSON files are the continuity artifacts for the next session.