Procedural Memory Knowledge Representation (PM-KR) Community Group
## Mission
The PM-KR Community Group develops standards for **procedural knowledge representation** that enable both humans and AI systems to consume the same canonical knowledge sources.
**Core insight (Milton Ponson, mandala graph theory):** Nothing is “wrong” with declarative approaches — they’re **necessary but insufficient**. PM-KR provides **procedural optimization given declarative foundation**.
**Boundary framework (Christoph Dorn):** Reality is not uniform, containing paradoxes and non-logical choices. PM-KR addresses hard/soft/blurred/broken boundaries at fractal levels, with structural transparency as the safety net for author accountability.
### Triple Foundation
PM-KR provides **procedural knowledge representation** where:
– **Declarative foundation** (Milton Ponson): Semantics, structure, mathematical rigor (mandala graph theory)
– **Procedural execution** (PM-KR): Runnable, renderable, multi-modal
– **Boundary framework** (Christoph Dorn): Hard/soft/blurred/broken boundaries at fractal levels
– **Structural transparency** (Christoph Dorn): Author accountability, safety net for AI systems
– **Reality modeling** (Christoph Dorn): Captures paradoxes, non-logical choices, ignorance (not uniform models)
**The result:** AI systems that are:
1. **Mathematically rigorous** (Milton’s mandala graph theory)
2. **Philosophically grounded** (Christoph’s boundary framework)
3. **Practically implementable** (reference implementations)
4. **Ethically accountable** (structural transparency = safety net)
—
## Problem Statement
Current knowledge representation systems suffer from massive duplication and fragmentation:
– The same knowledge is duplicated across fonts, embeddings, accessibility metadata, and visual renderings
– Human-readable and machine-readable formats diverge, requiring separate maintenance
– AI training data duplicates knowledge already available in structured forms
– Knowledge updates must be propagated across multiple systems manually
**Result:** Inefficiency, inconsistency, and unsustainability at scale.
—
## Approach: Declarative Foundation + Procedural Execution
**Declarative approaches** (RDF, OWL, JSON-LD) provide the **”know-what”** — foundational understanding (structure, relationships, semantics).
**PM-KR adds the “know-how”** — procedural execution layer that makes declarative knowledge **runnable, renderable, and multi-modal**.
**We’re not replacing RDF/OWL/JSON-LD — we’re adding the execution layer on top.**
### Core Principles
1. **Dual-Client Contract**: One canonical procedural source serves both humans (readable) and AI systems (executable)
2. **Compositional Architecture**: Knowledge units compose via symlink-style references (no duplication)
3. **Procedural Foundation**: Knowledge is executable programs (like TrueType fonts, mathematical formulas, physics simulations)
4. **Hyper-Modularity**: Atomic knowledge units that combine into complex structures
5. **Explicit Context Rules**: Context-dependent meanings handled via inspectable procedural rules (not implicit neural weights)
### Synergy with Declarative Standards
| **Approach** | **Role** | **Execution** | **Transparency** |
|————–|———-|—————|——————|
| **Declarative (RDF/OWL)** | **Foundation** (“know-what”) | ❌ No | ✅ Yes |
| **PM-KR (Procedural)** | **Execution layer** (“know-how”) | ✅ Yes | ✅ Yes |
| **Declarative + PM-KR** | **Complete system** | ✅ Yes | ✅ Yes |
—
## Scope
### In Scope
**Core Standards:**
– Procedural knowledge data models
– Composition semantics (symlink references, deduplication)
– Execution semantics (RPN-based procedural programs)
– Context rule semantics (handling context-dependent meanings)
– Conformance levels (minimal, extended, sovereign)
**Domains:**
– Mathematical knowledge (symbols, operators, formulas)
– Spatial knowledge (geometric primitives, transformations)
– Linguistic knowledge (characters, glyphs, typography)
– Educational knowledge (textbooks, curricula, multi-modal rendering)
– Game mechanics (rules, systems, procedural generation)
– Scientific knowledge (protocols, experiments, simulations)
**Relationships with W3C Technologies:**
– JSON-LD (structured data representation)
– RDF/OWL (semantic web integration — PM-KR adds execution layer)
– Verifiable Credentials (knowledge provenance)
– CBOR-LD (compression for efficient transmission)
### Out of Scope
– Proprietary AI training formats (focus: open, standardized)
– Natural language understanding (we provide structured knowledge inputs)
– Inference engines (we define knowledge representation, not reasoning systems)
—
## Deliverables (2026-2027)
### Specifications
1. **PM-KR Core Specification v1.0**
– Procedural knowledge data model
– Composition semantics
– Conformance levels
– **Target:** Q4 2026
2. **PM-KR Execution Semantics v1.0**
– RPN-based procedural programs
– Interpreter requirements
– Sandbox/security model
– **Target:** Q1 2027
3. **PM-KR Context Rules Specification v1.0**
– Context-dependent execution semantics
– Inheritance and override rules
– Multi-modal rendering patterns
– **Target:** Q2 2027
4. **PM-KR JSON-LD Profile v1.0**
– JSON-LD context definitions
– Vocabulary terms
– Canonical serialization rules
– **Target:** Q2 2027
### Reference Implementation
– **Knowledge3D** (Python/CUDA): Sovereign spatial procedural knowledge system with GPU execution
– Demonstrates: Hyper-modular architecture, 50-90% compression, dual-client contract
### Use Case Documentation
– **Education:** Procedural textbooks for human and AI tutors (MIT OpenCourseWare example)
– **Gaming:** Executable rulebooks for game masters and AI assistants (D&D SRD example)
– **Science:** Procedural experimental protocols (Nature Protocols example)
– **Accessibility:** Multi-modal knowledge rendering (visual, audio, tactile from ONE source)
– **AI Training:** Canonical knowledge sources (Wikipedia procedural KB — query, don’t duplicate)
—
## Current Members (18+ as of Feb 28, 2026)
**Chairs:**
– Daniel Campos Ramos (Founder, Knowledge3D Project)
– Milton Ponson (Co-Chair, Mathematical Foundations)
**Notable Organizations:**
– MIT Digital Credentials Consortium
– Huawei Technologies (W3C Advisory Board)
– Digital Bazaar (Manu Sporny, JSON-LD co-creator)
– LinkedIn (Knowledge Graphs)
– University of Brescia, Italy
– Indiana University
– Rensselaer Polytechnic Institute
– INRIA, France
– Cogsonomy
—
## Liaisons with Other W3C Groups
**Active Collaborations:**
– Verifiable Credentials WG: Knowledge provenance and attribution
– JSON-LD WG: Procedural JSON-LD extensions
– Spatial Data on the Web WG: Spatial knowledge representation
– Credentials CG: Academic credentials and structured educational knowledge
**Potential Future Liaisons:**
– Semantic Web Interest Group
– Web Machine Learning WG
– RDF-star WG (nested knowledge structures)
—
## Roadmap
**Q1 2026 (Current):**
– ✅ Community Group launched (Feb 20, 2026)
– ✅ 18+ members recruited
– ✅ Mission statement revised based on community feedback (v1.0 → v1.2)
– 🔄 Initial specification drafts (PM-KR Core v0.1)
**Q2 2026:**
– Publish PM-KR Core Specification v0.5 (draft for community review)
– Gather community feedback (use cases, implementations)
– Establish liaisons with related W3C WGs
– Expand reference implementations (JavaScript, Rust)
**Q3 2026:**
– Publish PM-KR Core Specification v1.0 (candidate)
– Host W3C TPAC breakout session
– Empirical validation studies (compression, performance, accuracy)
**Q4 2026:**
– Finalize PM-KR Core Specification v1.0
– Begin PM-KR Execution Semantics v1.0 draft
– Publish conformance test suite
**2027:**
– PM-KR Execution Semantics v1.0
– PM-KR Context Rules Specification v1.0
– PM-KR JSON-LD Profile v1.0
– Industry adoption (textbook publishers, game companies, AI platforms)
—
## Why PM-KR Matters Now
### Technical Impact
– **Standardize** procedural knowledge representation for AI systems
– **Reduce** knowledge duplication across systems (50-90% compression demonstrated)
– **Enable** dual-client knowledge sources (humans + AI from same source)
– **Complement** declarative standards (add execution layer to RDF/OWL/JSON-LD)
### Societal Impact
– **Education:** Accessible, multi-modal textbooks (visual, audio, tactile)
– **Sustainability:** Reduce AI’s carbon footprint via compression (no training data duplication)
– **Accessibility:** Knowledge rendered for diverse human needs (same source)
– **Reproducibility:** Scientific protocols as executable procedures
### Economic Impact
– **Efficiency:** Companies stop duplicating knowledge infrastructure
– **Innovation:** New applications enabled by compositional knowledge
– **Open Standards:** Prevent proprietary lock-in, foster ecosystem growth
—
## How to Participate
**Join:** https://www.w3.org/community/pm-kr/ (no W3C membership required)
**Contribute:**
– Review specifications in GitHub: https://github.com/danielcamposramos/Knowledge3D/tree/main/docs/vocabulary
– Propose use cases from your domain
– Build prototypes in your preferred language
– Participate in discussions: public-pm-kr@w3.org
—
## Contact
**Chairs:**
– Daniel Campos Ramos: capitain_jack@yahoo.com
– Milton Ponson: rwiciamsd@gmail.com
**Mailing List:** public-pm-kr@w3.org
**GitHub:** https://github.com/danielcamposramos/Knowledge3D
—
## Acknowledgments
PM-KR builds on decades of W3C work in semantic web, linked data, and web standards.
**Special thanks:**
– **Dave Raggett** (W3C veteran) — Critical feedback on declarative vs procedural positioning
– **Milton Ponson** (Co-Chair) — Mathematical insight: declarative foundation + procedural optimization synergy (mandala graph theory)
– **Christoph Dorn** (Systems Architect) — Boundary framework: hard/soft/blurred/broken boundaries, structural transparency as safety net for author accountability
– **Manu Sporny** (JSON-LD co-creator) — JSON-LD integration guidance
– **Tim Berners-Lee** — Linked Data principles
– All 18+ founding members who joined in the first week
—
**Version:** 1.3 (Triple Foundation: Milton + Christoph + Implementation)
**Last Updated:** February 28, 2026