RDF & SPARQL 1.2 Interoperability through Query Rewriting

RDF & SPARQL 1.2 Interoperability
through Query Rewriting

Jitse De Smet · Maxime Jakubowski

paper

The gap

RDF 1.2 adds triple terms: one native way to state metadata about a statement
Published data keeps its own reification pattern — rdf:Statement, n-ary, custom
A federation engine should bridge between different structures

Write SPARQL 1.2, then query the data you already have.

Mappings are just CONSTRUCT queries

1 Map the source onto the global schema

# rdf-reification → triple terms
CONSTRUCT { ?t rdf:reifies <<( ?s ?p ?o )>> }
WHERE { ?t a rdf:Statement ;
    rdf:subject ?s ; rdf:predicate ?p ;
    rdf:object ?o . }

# and pass everything else through
CONSTRUCT { ?s ?p ?o }
WHERE     { ?s ?p ?o }

2 Ask in SPARQL 1.2

SELECT * WHERE {
  ?t rdf:reifies <<( ?s rdf:type ?o )>> ;
     ex:confidence ?score .
  FILTER(?score > 0.8)
}

→

unfold

3 What the source actually gets

SELECT * WHERE {
  ?t a rdf:Statement ; rdf:subject ?s ;
     rdf:predicate rdf:type ; rdf:object ?o .
  ?t ex:confidence ?score .
  FILTER(?score > 0.8)
}

Extensiblenew schemes, no new code
Interpretableyou can read the mapping
Familiarit is just SPARQL
Portableany compliant engine

One rewriter, many mappings — all four ship as demo presets

reification

Standard reification

?st a rdf:Statement ;
  rdf:subject   ?s ;
  rdf:predicate ?p ;
  rdf:object    ?o .

n-ary

Statement nodes

?s  p:P31  ?st .
?st ps:P31 ?o .

beyond

Source selection

CONSTRUCT { ?s ?p ?o } WHERE {
  SERVICE <src> { ?s ?p ?o } }

beyond

Set-theoretic globals

{ …A… } UNION { …B… }
MINUS   { …C… }

Live demo Write mappings, pick sources, watch the rewrite — Traqula transforms, Comunica executes 2026-query-rewriting.demo.jitsedesmet.be

Beyond reification: general RDF-to-RDF mappings, formalised and optimised
Mappings we cannot yet write are input for SPARQL maintenance mode: CONSTRUCT GRAPH, triple-term templates, recursive views