A controlled, structured vocabulary that records preferred and non‑preferred terms and explicit semantic relations—equivalence (synonymy/use/used for), hierarchical (broader/narrower), and associative (related)—so that resources can be indexed and retrieved consistently and queries can be expanded or normalized across a collection.

Principle

By making preferred forms and inter‑term relations explicit, a thesaurus enforces consistent subject representation and enables both indexing and automated query expansion or normalization.

Demonstration

Situation: An indexer describes an article about heart attacks. Recognition: The thesaurus contains the preferred term “Myocardial Infarction” with an equivalence (UF) “Heart Attack” and broader term “Cardiovascular Diseases.” Action: The indexer assigns the preferred term; the search system maps user queries for “heart attack” to “Myocardial Infarction.” Consequence: Users searching common language retrieve the same set of items the indexer intended, increasing recall while preserving controlled indexing.

Misapplication

Treating an unstructured list of keywords as a thesaurus or assuming every ‘related term’ implies a hierarchical (is‑a) relationship. The semantic error is conflating presence of multiple words with defined semantic relations; without explicit equivalence/hierarchy definitions, query expansion and disambiguation will be incorrect.

Consequence

Correct use yields interoperable indexing, predictable retrieval behavior, and improved recall through controlled equivalence and expansion. Misuse or poorly defined relations produce inconsistent indexing, false synonymy, over‑expansion of queries, and degraded precision.

Reversal

Where domain concepts require formal logical inference, cardinality constraints, or richly axiomatized relationships (e.g., ontologies with class axioms), a traditional thesaurus may be insufficient; conversely, in rapidly changing or very small domains, lightweight vocabularies or dynamic tagging may be preferable.

Boundary

Clearly within: a controlled vocabulary with documented preferred terms, non‑preferred forms, and defined broader/narrower/related relations (e.g., SKOS thesaurus). Boundary case: a curated list that records preferred labels and synonyms but lacks explicit hierarchical/associative relations. Clearly outside: unordered user tags or classification notations that provide no term equivalence or relational structure.

Semantic Tension

Standardization (fixed preferred terms and relations) ↔ Flexibility (user language and evolving terminology): tighter control improves interoperability but can reduce responsiveness to new or localized language.

Synthesis

A thesaurus is both a normative lexical map (linking variants to preferred terms) and a relational network; its value lies in explicit, governed relations that translate human conceptual choices into machine‑actionable controls for indexing and retrieval.