Keep a lesson, diagram or model in one place so people and AI can refer to the same knowledge. Knowledge3D explores shared knowledge in 3D spaces for researchers, developers and curious readers, including people without a technical background.
Status: the code is paused. Work now focuses on the specifications and reports for PM-KR, a W3C community group studying how people and AI share knowledge. These project drafts are proposals for discussion, not adopted W3C standards.
Start here: read the explainer for the problem, the proposed first step and the questions still open.
The idea is to store knowledge once as procedures, instructions that describe how to make or do something, with people and AI reading it in the same place.
A font file is the working precedent. Glyphs describe the shapes of letters; OpenType shaping rules combine and position them; Unicode identifies the characters those shapes represent. The person sees a letter and software can identify it. Character identity is a starting point for meaning; a font does not explain a whole concept.
The next proposed deliverable is a small “font file” repository for knowledge: a procedure for form, a SKOS term for meaning, and PROV records for origin and time. SKOS names concepts and their relationships; PROV records where information came from. This would sit alongside RDF, OWL and JSON-LD, the web's ways to describe and exchange linked knowledge. Daniel's 28 October proposal and clarification explain the font precedent and why each subject needs a useful stopping point for breaking knowledge into parts.
- Standards: the document index labels drafts, discussion inputs and superseded texts.
- Researchers: the use cases and open questions identify what needs testing.
- Implementers: the proposed approach connects the model, conformance drafts and exchange guide.
- Take part: the PM-KR Community Group is open to anyone with a W3C account, subject to its participation terms.
Daniel Campos Ramos directs Knowledge3D, with AI assistance from Claude, Codex and the contributors credited in Attributions. The project's influences are recorded separately. Code is under Apache 2.0; the white paper is under CC BY 4.0. Cite the project using CITATION.cff and DOI: 10.5281/zenodo.23308993.