Mathematical and computational foundations

The argument of this section

The first section asked what it takes for two people to mean the same thing. This one asks what it takes for a machine to participate. The answer of the twentieth and twenty-first centuries was formalization: write concepts down as sets, classes, and relations, state the rules that govern them, and let software check whether a new statement is consistent with the old ones. That made new things possible, such as finding every record that falls under a concept without listing them by hand, or detecting that two definitions contradict each other. It also carried assumptions. A representation only knows what was put into it, an inference is only as good as its axioms, and a graph that stores facts is not the same as a logic that reasons over them.

This section walks through that machinery in order: controlled vocabularies, graphs, description logics, the semantic web standards that give them a common syntax, and the terminology servers through which clinical systems use them. SNOMED CT is the most important clinical case, not the whole program. Examples come from engineering, political science, biology, history, and medicine, because the ideas are general.

How to read it

Two essays frame the section. When meaning becomes computable introduces sets, classes, instances, predicates, and constraints, compares a plain Knowledge graph with a formal ontology, and demonstrates one inference and one counterexample on a reproducible toy model, so that the reader sees exactly which assumption the conclusion depends on. One concept across many systems follows a synthetic observation as it crosses between systems and explains what identifiers, code systems, value sets, and concept maps do when meaning has to travel.

Six monographs give the reference depth. Controlled vocabularies and Knowledge graphs cover the two most common and most often confused tools. Description logics explains the logics behind OWL and why decidability matters to anyone who wants an answer in finite time. Semantic web standards treats RDF, OWL, and their companions as a shared syntax rather than a philosophy. SNOMED CT and the NucLex niche examines how SNOMED CT is built and where nuclear medicine falls outside its coverage. FHIR terminology resources explains how FHIR packages a Code system, a Value set, and a Concept map so that a Terminology server can serve them. The lexicon entries, from Description logic and Reasoning onward, are short and link back to the monographs for argument.

Why it matters here

A nuclear medicine physician meets this machinery every time an order, a report, or a dose record moves between systems. A scientist pooling imaging and laboratory data across sites depends on it whether or not the tool makes it visible. A radiochemist who wants a labeled compound recognized as the same substance in a regulatory file, a pharmacy system, and a research database is asking a question only identifiers and mappings can answer. Understanding what formalization can and cannot do is the difference between trusting a system and being able to say why. This section describes established methods and standards; NucLex itself operates no terminology server, reasoner, or mapping service in W0, and its essays and monographs are drafts and briefs awaiting review.