Tools and roadmap

The argument of this section

NucLex is a publication first and a platform later. The project's principle is to build the living prototype before any bureaucracy, and to stop and reflect before expanding. What you are reading is the prototype. Every stage described after it is a plan, a question, or a research direction, not a delivered capability.

Two labels must be kept apart because the project's history used "Version 1" for both. W0 is this website: an information architecture, writing briefs, draft essays, monographs, and concept entries, and a roadmap, maintained as versioned content files. It has no terminology endpoints, reasoner, mapping service, or atlas. Technical V1 is the SNOMED CT-anchored terminology platform proposed in the founding discussion: a nuclear medicine extension of SNOMED CT with a terminology server, a versioned ontology source, and published code systems and value sets. It has not been built.

The later stages are planned in sequence. V1A would add an agentic layer above V1 in which agents survey literature and propose terms, definitions, relations, and mappings, each packaged with evidence; agents propose and humans approve. V2 would add the second and third pillars: a curated registry of preferred external authorities and a composer that builds compound concepts with provenance. V3 would add standards-style governance, from proposal and triage through expert review, public comment, approval, publication, and deprecation. V4 would make NucLex a public service through documented APIs, agent interfaces, and bulk releases. V4 plus would add interactional knowledge for models: entities, interactions, parameters, and evidence, each with context, units, uncertainty, and provenance. No dates, budgets, or staffing commitments are attached to any stage.

How to read it

The three pillars of NucLex explains the three functions meant to operate inside the nuclear medicine niche: reuse of SNOMED CT and other authorities, curation of which source to trust for which kind of Concept, and composition of what is missing, walked through on a hypothetical compound concept with its Provenance and review. From living publication to interactional knowledge takes the roadmap from W0 to V4 plus and asks what evidence should justify each step.

Three monographs carry the technical argument. Interactional ontology develops the distinction the founding discussion insisted on: the properties of molecule A, of molecule B, and of how A behaves toward B are different kinds of knowledge, and the third is the least well represented. Human review of machine proposals examines why Human review is a requirement of legitimacy rather than a bottleneck to optimize away. Model-ready knowledge exchange asks what a model would need from a knowledge source. The lexicon entries Interactional ontology, Interaction, Parameter, Uncertainty, and Context define the final stage's vocabulary in short form.

Why it matters here

A nuclear medicine physician should be able to tell from this section what NucLex can do today (explain, orient, define) and what it cannot yet do (serve codes, resolve identifiers, feed a model). A scientist or radiochemist should see where their expertise would enter: as a reviewer of proposed concepts, a curator of a preferred source, or a contributor of the parameters and interactions a later stage would need. Progress is to be judged by what works, not by what is described. Everything here is a draft under editorial review.