eyeprolog

RDF 1.2 triple terms

Can a statement be talked about without being asserted?

Alice claims that Bob knows Carol. RDF 1.2 can record that claim as a triple term: a triple used as a value inside another triple. EyeProlog reads it as an ordinary nested Prolog term and lifts the quoted relationship out into a result of its own.

Run it in the playground · Source

The question

What relationship is inside Alice’s claim?

This is one of a family of RDF 1.2 examples. RDF data arrives as rdf(Subject, Predicate, Object, Graph) facts, as produced by rdf-prolog-interchange; ordinary rules derive new results in the same shape, and write_results/0 prints them as rdf/4 facts that convert straight back to RDF.

The data

rdf(iri('https://example.org/alice'), iri('https://example.org/claims'), triple(iri('https://example.org/bob'), iri('https://example.org/knows'), iri('https://example.org/carol')), default_graph).

A triple term is a whole statement used as the object of another statement. Here the object of claims is triple(bob, knows, carol). The data does not say that Bob knows Carol; it says that Alice claims so.

The rule

result_rdf(S, iri('https://example.org/knows'), O, G) :-
  rdf(
    _,
    iri('https://example.org/claims'),
    triple(S, iri('https://example.org/knows'), O),
    G
  ).

Unification looks inside the triple term: the pattern triple(S, knows, O) matches the quoted statement and binds its subject and object. Projecting a claim into a plain relationship is a decision this rule makes explicitly; nothing in RDF does it for you.

The answer

eyeprolog examples/rdf12-triple-term.pl

result_rdf(iri('https://example.org/bob'), iri('https://example.org/knows'), iri('https://example.org/carol'), default_graph).

One relationship, recovered from inside the claim.

The reason

eyeprolog --proof examples/rdf12-triple-term.pl

step(result_rdf(iri('https://example.org/bob'), iri('https://example.org/knows'), iri('https://example.org/carol'), default_graph),
     rule(2),
     ['S' = iri('https://example.org/bob'),
      'O' = iri('https://example.org/carol'),
      'G' = default_graph],
     [rdf(iri('https://example.org/alice'), iri('https://example.org/claims'), triple(iri('https://example.org/bob'), iri('https://example.org/knows'), iri('https://example.org/carol')), default_graph)]).

The result rests on exactly one source triple: Alice’s claim. An auditor can see that the asserted relationship came from a claim, and whose claim it was.

The check

condition('C1', resolution, ok, 2).
condition('C4', coverage, ok, 2).
condition('C7', relevance, ok, 3).
trusted(0).
verdict(checked).

Both steps replay against the program and nothing is taken on trust.

Try it

Change knows to likes inside the triple term. The run prints nothing: the rule only projects knows claims.
Add a second claim by another person and require that two different people claim the same relationship before it is projected.
Compare with the related decks: nested triple terms, annotations, and triple terms in a named graph.

Learn more in The Art of EyeProlog, Chapter 9.

This site is open source. Improve this page.