Data properties

Internal structure of knowledge subjects

In databases and programming, data properties (DP) are often called attributes. They describe internal characteristics of knowledge subjects, whereas object properties relate subjects to external objects. In conceptual-space style accounts, qualities such as weight, colour, or temperature define dimensions for comparison; they often pair with physical units (see the canonical Physical quantities branch on taoke.de).

The treatise adds several DP types that affect how instantiation works, including transparent and universals data properties—important for multi-level modeling simplifications. Naming conventions (DP naming on taoke.de) let readers infer the DP type from the identifier; DPs can be grouped in hierarchies analogous to class hierarchies. Pairs of (DP, value from its value set) are called features and appear in formal concept analysis cross-tables. [AlHe2020]; DL concrete domains in [BaMc2003].

Rules for DP instantiations depend on DP type and class level (particulars vs. schema layer). In short: use particulars DPs over atomic types when values attach to individuals; use transparent DPs when values should propagate to subclasses and particulars (value assignment propagation); use universals DPs for schema-layer class metadata; use meta DPs for annotations. Transparent DPs can avoid redundant parallel class/particular pairs via powertype-style patterns (see [BeHu2021]).

A data property definition can be written as a triple (^C, .dp, :adt) (short form) or linked to AKE patterns; short and long forms can be translated with SPARQL-style patterns [SPARQL] as on the canonical page.

The canonical Nuclides example shows subclasses of ^Nuclide with transparent DPs such as .∆Neutrons and .∆Protons: values can be asserted at class/meta-class level and need not be repeated for every particular. A definition triple uses :Integer as object; a value assertion uses a plain literal—syntax distinguishes definition from assertion.

Literal fields in triples and rule metadata play the same practical role: keep datatypes explicit and literals normalized so rules and exports stay predictable.