Why the Triple Is the Wrong Grain — An RDF Rebuttal

PREFIX : <https://medium.com/@JamesStakelum/why-the-triple-is-the-wrong-grain-1cad99e136f4#>
CONSTRUCT { :beethovenNinthComposition ?p ?o }
FROM <https://linkeddata.uriburner.com/DAV/demos/daas/why-the-triple-is-the-wrong-grain-rebuttal-demo-data-claude_code-1.ttl>
WHERE {
  :beethovenNinthComposition ?p ?o .
}

Verified live against the loaded demo dataset. Returns the whole schema:CreateAction event (agent, result, location, time) in one query — nothing about it was 'shattered'. If you load this event into its own named graph (see :c1 :namedGraphExample), swap the FROM/subject for GRAPH <that-iri> { ?event ?p ?o } for atomic graph-scoped retrieval.

PREFIX : <https://medium.com/@JamesStakelum/why-the-triple-is-the-wrong-grain-1cad99e136f4#>
PREFIX rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>
PREFIX prov: <http://www.w3.org/ns/prov#>
SELECT DISTINCT ?claim ?place ?source
FROM <https://linkeddata.uriburner.com/DAV/demos/daas/why-the-triple-is-the-wrong-grain-rebuttal-demo-data-claude_code-1.ttl>
WHERE {
  ?claim rdf:predicate :birthPlace ; rdf:object ?place ; prov:wasAttributedTo ?source .
}

Verified live: returns BOTH :claimBirthPlaceBonn and :claimBirthPlaceKoblenz, each with its own source attribution — never a merged/averaged third place. Loading each claim into its own named graph (per :c7 :namedGraphExample) makes the same point via GRAPH ?g { :Beethoven :birthPlace ?place } instead of reification.

PREFIX : <https://medium.com/@JamesStakelum/why-the-triple-is-the-wrong-grain-1cad99e136f4#>
PREFIX rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>
SELECT DISTINCT ?claim ?subject ?predicate ?object ?tier
FROM <https://linkeddata.uriburner.com/DAV/demos/daas/why-the-triple-is-the-wrong-grain-rebuttal-demo-data-claude_code-1.ttl>
WHERE {
  ?claim a rdf:Statement ;
      rdf:subject ?subject ; rdf:predicate ?predicate ; rdf:object ?object ;
      :epistemicStatus ?tier .
}

Verified live: separates the documented birthplace fact (:DocumentedFact) from the inferred birth year (:Inferred) by claim-level :epistemicStatus — the tier a FuXi N3 rule would otherwise materialize from named-graph membership onto each triple.

PREFIX : <https://medium.com/@JamesStakelum/why-the-triple-is-the-wrong-grain-1cad99e136f4#>
SELECT DISTINCT ?person ?relation ?value
FROM <https://linkeddata.uriburner.com/DAV/demos/daas/why-the-triple-is-the-wrong-grain-rebuttal-demo-data-claude_code-1.ttl>
WHERE {
  ?person ?relation ?value .
  FILTER(?relation = :hasCitizenship)
}

Verified live: Alice's citizenship is declared via :hasCitizenship, a rdfs:subPropertyOf :hasRelationalStatus — new relational predicates stay open-vocabulary while an RDFS reasoner keeps them bounded under the small superproperty set. Under RDFS entailment, ?relation = :hasRelationalStatus also matches via subproperty inference.

DEFINE input:same-as "yes"
PREFIX : <https://medium.com/@JamesStakelum/why-the-triple-is-the-wrong-grain-1cad99e136f4#>
ASK { :Beethoven a <http://dbpedia.org/ontology/Person> }

Verified live both ways: WITHOUT the DEFINE input:same-as pragma this ASK returns false (:Beethoven carries no direct rdf:type to dbo:Person, only owl:sameAs to the DBpedia resource); WITH it, true — deterministic OWL2 entailment via Virtuoso's built-in same-as reasoning (no named ruleset needed), not a probabilistic similarity score. Deliberately has no FROM/GRAPH restriction: the entailment bridges our locally-asserted owl:sameAs triple with DBpedia's own rdf:type assertion, which lives in a different graph elsewhere in the store. See :demoQ9 for a SELECT variant showing the full scale of what gets entailed (71 types, not just one boolean).

PREFIX : <https://medium.com/@JamesStakelum/why-the-triple-is-the-wrong-grain-1cad99e136f4#>
PREFIX prov: <http://www.w3.org/ns/prov#>
SELECT DISTINCT ?first ?firstTime ?second ?secondTime
FROM <https://linkeddata.uriburner.com/DAV/demos/daas/why-the-triple-is-the-wrong-grain-rebuttal-demo-data-claude_code-1.ttl>
WHERE {
  ?first :precedes ?second .
  ?first prov:endedAtTime ?firstTime .
  ?second prov:startedAtTime ?secondTime .
}

Verified live: explicit :precedes plus xsd:dateTime literals recover chronology (poisoning before death) regardless of narrative order.

PREFIX : <https://medium.com/@JamesStakelum/why-the-triple-is-the-wrong-grain-1cad99e136f4#>
PREFIX rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>
SELECT DISTINCT ?statement ?subject ?predicate ?object ?speechAct
FROM <https://linkeddata.uriburner.com/DAV/demos/daas/why-the-triple-is-the-wrong-grain-rebuttal-demo-data-claude_code-1.ttl>
WHERE {
  ?statement a rdf:Statement ;
      rdf:subject ?subject ; rdf:predicate ?predicate ; rdf:object ?object ;
      :speechAct ?speechAct .
}

Verified live: the reified statement (the defendant's utterance, and separately Beethoven's composition claim) carries its own :speechAct — belief-about-the-claim and the underlying triple are two distinct, independently queryable resources. RDF-star's <<...>> syntax expresses the same distinction more compactly, and already does on Virtuoso Open Source Edition; the commercial edition behind this document's live URIBurner endpoint will support these RDF release enhancements in an upcoming release, so reification is the verified mechanism used here.

PREFIX : <https://medium.com/@JamesStakelum/why-the-triple-is-the-wrong-grain-1cad99e136f4#>
PREFIX rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>
PREFIX prov: <http://www.w3.org/ns/prov#>
SELECT DISTINCT ?claim ?subject ?predicate ?object ?attributedTo ?generatedAt ?hash
FROM <https://linkeddata.uriburner.com/DAV/demos/daas/why-the-triple-is-the-wrong-grain-rebuttal-demo-data-claude_code-1.ttl>
WHERE {
  BIND(:claim42 AS ?claim)
  ?claim a rdf:Statement ;
      rdf:subject ?subject ; rdf:predicate ?predicate ; rdf:object ?object ;
      prov:wasAttributedTo ?attributedTo .
  OPTIONAL { ?claim prov:generatedAtTime ?generatedAt }
  OPTIONAL { ?claim :contentHash ?hash }
}

Verified live (exactly one row): returns :claim42's identity (its own IRI), lexical material (rdf:subject/predicate/object), provenance (prov:wasAttributedTo/generatedAtTime), and integrity check (:contentHash) in one query — the four SGF 'admission context' fields, already RDF-native. Other reified claims in the dataset (:claimComposedNinth, :claimBirthPlaceBonn/Koblenz) also carry prov:wasAttributedTo but not generatedAtTime/contentHash — drop the BIND to compare all of them side by side. Packaging this claim as its own signed named graph (per :c11 :namedGraphExample) moves identity from the claim IRI to the graph IRI, same four fields.

DEFINE input:same-as "yes"
PREFIX : <https://medium.com/@JamesStakelum/why-the-triple-is-the-wrong-grain-1cad99e136f4#>
SELECT DISTINCT ?type
WHERE { :Beethoven a ?type }

Verified live: without the DEFINE input:same-as pragma this returns 2 rows (our own asserted schema:Person and :Composer); with it, 71 rows — the full type hierarchy (YAGO categories, schema.org, foaf, dbo, ...) that DBpedia's Ludwig_van_Beethoven resource carries, all entailed transitively through the single owl:sameAs link. A far more vivid demonstration of what OWL2 entailment actually pulls in than :demoQ5's bare true/false ASK — same underlying mechanism, richer evidence. See :demoQ11 for the same query with the pragma commented out, for a direct one-click A/B comparison.

# DEFINE input:same-as "yes"   -- uncomment this line to enable reasoning and compare results
PREFIX : <https://medium.com/@JamesStakelum/why-the-triple-is-the-wrong-grain-1cad99e136f4#>
ASK { :Beethoven a <http://dbpedia.org/ontology/Person> }

Verified live: identical to :demoQ5 except the DEFINE input:same-as pragma is commented out (a plain SPARQL '#' comment, inert until you delete the leading '# '), so this returns false — no direct rdf:type assertion, only owl:sameAs. Uncomment that one line and re-run to see it flip to true, exactly as :demoQ5 does — the fastest way to see OWL2 entailment turn on and off for yourself.

# DEFINE input:same-as "yes"   -- uncomment this line to enable reasoning and compare results
PREFIX : <https://medium.com/@JamesStakelum/why-the-triple-is-the-wrong-grain-1cad99e136f4#>
SELECT DISTINCT ?type
WHERE { :Beethoven a ?type }

Verified live: identical to :demoQ9 except the DEFINE input:same-as pragma is commented out, so this returns only 2 rows (our own asserted schema:Person and :Composer) instead of 71. Uncomment that one line and re-run to watch the full DBpedia type hierarchy — YAGO categories, schema.org, foaf, dbo — flood in through the single owl:sameAs link, exactly as :demoQ9 does.