Sometimes you want to express facts about facts - say "the triple 'trumpet[hasSize -> small]' hasEvidence high.

Similarly to RDF this cannot be done directly, for an OBL molecule has no ID by itself.

The most appropriate modelling pattern for this problem is called reification: We make a fact "Subject [ property -> Object]" being the subject of another fact by itself.

The fact we want to talk about:

erniesFireEngine hasColor red

ex#erniesFireEngine [ ex#hasColor -> "red" ] .

Partial reification

Idea: One table "property(xyz)" for each single property

ex#property(hasSize) :: ex#Properties .

turtle: ex#property(hasSize) rdfs:subClassOf ex#Properties .

ex#property(hasColor) :: ex#Properties .

turtle: ex#property(hasColor) rdfs:subClassOf ex#Properties .

ex#'fact (erniesFireEngine,hasColor, red)' : ex#property(hasColor) .

turtle: ex#'fact (erniesFireEngine,hasColor, red)' rdf:type ex#property(hasColor) .

ex#'fact (erniesFireEngine,hasColor, red)' [ ex#hasSubject -> ex#erniesFireEngine ] .

ex#'fact (erniesFireEngine,hasColor, red)' [ ex#hasObject -> ex#red ] .

ex#'fact (erniesFireEngine,hasColor, red)' [ ex#hasEvidence -> ex#high ] .

Full reification

Idea: One single table "ReifiedFacts" for all properties together

Problem: inferencing optimization will be knocked out rather fundamentally.

ex#hasSize : ex#Property .

turtle: ex#hasSize rdf:type ex#Property .

ex#hasColor : ex#Property .

turtle: ex#hasColor rdf:type ex#Property .

ex#'fact (erniesFireEngine,hasColor, red)' : ex#ReifiedFacts .

turtle: ex#'fact (erniesFireEngine,hasColor, red)' rdf:type ex#ReifiedFacts .

ex#'fact (erniesFireEngine,hasColor, red)' [ ex#hasSubject -> ex#erniesFireEngine ] .

ex#'fact (erniesFireEngine,hasColor, red)' [ ex#hasProperty -> ex#hasColor ] .

ex#'fact (erniesFireEngine,hasColor, red)' [ ex#hasObject -> ex#red ] .

ex#'fact (erniesFireEngine,hasColor, red)' [ ex#hasEvidence -> ex#high ] .