A Common Representation for Cross-Linguistic Meaning: Computational Architecture for Sub-Graph Matching Search on Ultrametric Semantic Trees
Charter Date: 2026-05-22 Project Lead: Rowan Brad Quni-Gudzinas Program: Ultrametricity Phase: P2 Execution
Most global digital infrastructure — search engines, databases, large language models — is built on the structural assumptions of English, an isolating language where words are separate, minimally inflected units. These assumptions systematically fail for polysynthetic languages, where a single word can encode what English requires an entire clause to express. The result is a digital divide: 10-20 million polysynthetic language speakers cannot effectively search, index, or retrieve content in their own languages.
Model meaning as nested ultrametric trees (nested semantic graphs) rather than flat token sequences. In this representation, the English sentence, the Mohawk word-sentence, and the Turkish agglutinative chain all map to the same underlying structure. A search architecture built on sub-graph matching across these trees would be genuinely language-neutral.
This project is a computational implementation sequel to the published conceptual paper "Few Become One: Polysynthetic Communication and the Ultrametric Architecture of Language" (DOI: 10.5281/zenodo.20328374, 2026-05-22). Where "Few Become One" argues that language is an ultrametric tree, this project specifies how to build search, ranking, and matching systems on that tree.
| Deliverable | Description |
|---|---|
| Internal Literature Review (0.1.md) | Comprehensive due diligence across the research corpus |
| Formal Definitions (0.2.md, 0.2.py) | Mathematical grounding with Python verification |
| Search Architecture Spec (0.3.md) | Sub-graph matching, ranking, Q-PNA integration |
| Cross-Linguistic Examples (0.4.md) | English/Turkish/Mohawk isomorphic NSTs |
| Python Prototype (0.5.py) | Parser, matcher, ranking engine |
| Computational Pathway (0.6.md) | Full pipeline architecture and feasibility assessment |
| Reader Testing & Publication | Blind validation and polish for release |
- Full-scale production search engine (BACKLOG P3)
- Training neural semantic parsers (BACKLOG P3)
- Real corpus validation of Language-Info-Architecture entropy values
- Native speaker verification of Mohawk/Iroquoian examples
- Large-scale cross-linguistic corpus collection (BACKLOG P2)
- This project does NOT argue that language is an ultrametric tree — that argument is in "Few Become One"
- This project does NOT implement the Q-PNA neural architecture — that specification is published separately
- This project does NOT provide a production-ready search engine — it provides the architecture specification and prototype
| Criterion | Measurement |
|---|---|
| Formal definitions are complete and Python-verified | All ultrametric checks pass (0 violations) |
| Search architecture is fully specified | Three-tier matching, ranking algorithm, Q-PNA pipeline documented |
| Cross-linguistic isomorphism is demonstrated | English, Turkish, Mohawk produce identical NSTs |
| Python prototype executes correctly | Parser → tree → matcher → ranking pipeline functional |
| Reader testing passes | No BLOCKING or MAJOR issues remaining |
| Publication-ready document produced | Passes Publication Language Gate (§11.7), has DOI |
| Constraint | Value |
|---|---|
| Write sandbox | G:\My Drive\projects\nested-semantic-graph\ |
| Branch | feature/initial-setup |
| Python | Standard library only (unless documented) |
| Math formatting | LaTeX/MathJax only (§6, Rule 6) |
| Citations | External source files or [LLM-INFERRED] label |
| Publication | User approval required before Obsidian\releases\ move |
| Stakeholder | Role |
|---|---|
| Rowan Brad Quni-Gudzinas | Author, domain expert, approver |
| Projects Agent | Primary executor |
| REVIEWER Subagent | Blind reader testing |
| QWAV Agent | Portfolio-level coordination (informed, not directing) |
| Dependency | Type | Status |
|---|---|---|
| Few Become One (DOI: 10.5281/zenodo.20328374) | Conceptual foundation | Published |
| Q-PNA (DOI: 10.5281/zenodo.20287742) | Neural architecture | Published |
| Language-Info-Architecture (DOI: 10.5281/zenodo.20137616) | Quantitative grounding | Published |
| Tree Cophenetic (DOI: 10.5281/zenodo.20213043) | Mathematical formalization | Published |
| ultrametric-ai-poc (GitHub) | Reference implementation | Published |
See RISK-REGISTER.md for complete risk tracking.
| Role | Name | Date | Status |
|---|---|---|---|
| Project Lead | Rowan Brad Quni-Gudzinas | 2026-05-22 | Initiated |
| Charter Accepted | — | — | Pending formal acceptance |
Generated from PROJECT-CHARTER-TEMPLATE.md v1.0.