Home > Tools > A–Z > Category theory

Category theory

Category theory held as a tool — an instrument for relational analysis. For the field on its own terms, see the positioning subject page. For where category theory meets the seed, see when category theory and the seed meet.

Three features that carry the weight

• Structural comparison. Functors map one system of relations onto another, preserving composition. The question shifts from “are these the same?” to “is there a structure-preserving map between them?”
• Relational characterisation. The Yoneda Lemma — an object is fully characterised by its relationships — licenses treating subjects and languages by their relational profiles rather than their interior content.
• Level-climbing. Higher categories articulate relations between relations, indefinitely up. Each level brings finer expression about structure already present in the base — the relational power is in the starting primitives, not added by climbing.

SPLectrum’s use

Category theory is the tool SPLectrum uses to look across languages. Mapping one language onto another is a functor. Comparing two such mappings is a natural transformation. The apparatus scales: from two languages to a web of them, from simple translation to translation-of-translations.

• Cross-language analysis. Every language (natural, formal, software, domain-specific) is a category. Mappings between them are functors. CT gives the shape of what is preserved and what is lost in translation.
• Seed formalisation. P0 and P1 are CT’s own starting primitives in philosophical form. Given those two, CT implements the structural dynamics of the remaining principles — the meaning stays in the seed.
• The float principle. CT reasons about relations without committing to what the objects are. This is what makes it usable across SPLectrum’s full range — from philosophical traditions to software protocols.

Prior art

SPLectrum’s use of CT sits within an existing tradition of applying category theory to language and to translation between formalisms.

Lambek — categorical grammar. Syntax as a pregroup, where grammatical types compose by adjunction. The first systematic use of CT as a tool for language structure.

DisCoCat — distributional compositional categorical models. Coecke, Sadrzadeh, Clark. Combines distributional semantics (meaning from context) with categorical composition (meaning from structure). The sentence is a morphism; meaning composes categorically.

Institutions (Goguen, Burstall) — a category-theoretic framework for relating logical systems. Each institution is a logical system with its own signatures, sentences, and satisfaction relation. Translations between institutions are functors. A framework for comparing formalisms without reducing them to a common one.