ELIS  –  Mul*media  Lab
Reasoned
SPARQL
Sam Coppens, Miel Vander Sande, Ruben Verborgh,
Erik Mannens, Rik Van de Walle,
SPARQL
JACK BAUER INTERROGATION TECHNIQUE
When asking politely just isn’t enough
Reasoning on
distributed data
Reasoned
SPARQL
ELIS  –  Mul*media  Lab
Goal: Reasoning as a Service
Smart SPARQL agents
outsource reasoning to appropriate infrastructure
(client-side, server-side, or third party)
Client
Server /
Data
Provider
Reasoning OWL QLOWL RL / EL
ELIS  –  Mul*media  Lab
Smart SPARQL Agent
Goal: Reasoning as a Service
Client
Server /
Data
Provider
Reasoning
Reasoning
Grid
Goal: Reasoning as a Service
Smart SPARQL agents
outsource reasoning to appropriate infrastructure
(client-side, server-side, or third party)
ELIS  –  Mul*media  Lab
Distributed Reasoning (LarKC)
Identification
Selection
Transformation
Reasoning
Decision
REASON
<http://test.com/rules.n3>
OVER {
?s ?p ?o
}
WHERE {
?s ?p ?o
}
Solution: ‘Reason’ Query Form
ELIS  –  Mul*media  Lab
SPARQL 1.1: Support for entailment regimes
(RDFS, OWL) by means of BGP matching
Data provider decides inference rules support
With reasoned SPARQL, the data consumer
choses the inference rules for reasoning
ELIS  –  Mul*media  Lab
Nested Queries
SELECT ?child
WHERE {
:Jenna :child ?child .
{
REASON {
{ ?x :parent ?y } => { ?y :child ?x } .
}
OVER {
?s :parent ?o .
}
WHERE {
?s a :Person; :parent ?o .
}
}
}
ELIS  –  Mul*media  Lab
•
Workload Balancing
REASON {
{ ?x foaf:knows ?y } => { ?y foaf:knows ?x } .
}
OVER {
:Jenna foaf:knows ?person .
}
WHERE{
{
SERVICE <http://example.org/sparql> {
:Jenna foaf:knows ?person .
} } UNION {
SERVICE <http://example2.org/sparql> {
:Jenna foaf:knows ?person .
} }
}
ELIS  –  Mul*media  Lab
Pitfalls
Incomplete reasoning / errors
like SPARQL
Server-side vs. client-side reasoning
probably OWL QL reasoning
server-side BGP matching
OWL RL and EL will happen client-side
ELIS  –  Mul*media  Lab
Distributed reasoning on top of distributed querying
Reasoning as a service
Use Cases
End user
Distributed
Reasoning
Framework
Endpoints
Endpoints
Endpoints
Client
External
Reasoner
Endpoint
ELIS  –  Mul*media  Lab
Reasoning on
distributed data
SPARQL
Reasoned
SPARQL
ELIS  –  Mul*media  Lab
ELIS  –  Mul*media  Lab
Sam Coppens, Miel Vander Sande, Ruben Verborgh,
Erik Mannens, Rik Van de Walle,
Reasoned
SPARQL
Using SPARQL agents to obtain answers
through distribution and reasoning

Reasoned SPARQL

  • 1.
    ELIS  –  Mul*media Lab Reasoned SPARQL Sam Coppens, Miel Vander Sande, Ruben Verborgh, Erik Mannens, Rik Van de Walle,
  • 3.
  • 4.
    JACK BAUER INTERROGATIONTECHNIQUE When asking politely just isn’t enough Reasoning on distributed data
  • 5.
  • 7.
    ELIS  –  Mul*media Lab Goal: Reasoning as a Service Smart SPARQL agents outsource reasoning to appropriate infrastructure (client-side, server-side, or third party) Client Server / Data Provider Reasoning OWL QLOWL RL / EL
  • 8.
    ELIS  –  Mul*media Lab Smart SPARQL Agent Goal: Reasoning as a Service Client Server / Data Provider Reasoning Reasoning Grid Goal: Reasoning as a Service Smart SPARQL agents outsource reasoning to appropriate infrastructure (client-side, server-side, or third party)
  • 9.
    ELIS  –  Mul*media Lab Distributed Reasoning (LarKC) Identification Selection Transformation Reasoning Decision REASON <http://test.com/rules.n3> OVER { ?s ?p ?o } WHERE { ?s ?p ?o } Solution: ‘Reason’ Query Form
  • 10.
    ELIS  –  Mul*media Lab SPARQL 1.1: Support for entailment regimes (RDFS, OWL) by means of BGP matching Data provider decides inference rules support With reasoned SPARQL, the data consumer choses the inference rules for reasoning
  • 11.
    ELIS  –  Mul*media Lab Nested Queries SELECT ?child WHERE { :Jenna :child ?child . { REASON { { ?x :parent ?y } => { ?y :child ?x } . } OVER { ?s :parent ?o . } WHERE { ?s a :Person; :parent ?o . } } }
  • 12.
    ELIS  –  Mul*media Lab • Workload Balancing REASON { { ?x foaf:knows ?y } => { ?y foaf:knows ?x } . } OVER { :Jenna foaf:knows ?person . } WHERE{ { SERVICE <http://example.org/sparql> { :Jenna foaf:knows ?person . } } UNION { SERVICE <http://example2.org/sparql> { :Jenna foaf:knows ?person . } } }
  • 13.
    ELIS  –  Mul*media Lab Pitfalls Incomplete reasoning / errors like SPARQL Server-side vs. client-side reasoning probably OWL QL reasoning server-side BGP matching OWL RL and EL will happen client-side
  • 14.
    ELIS  –  Mul*media Lab Distributed reasoning on top of distributed querying Reasoning as a service Use Cases End user Distributed Reasoning Framework Endpoints Endpoints Endpoints Client External Reasoner Endpoint
  • 15.
    ELIS  –  Mul*media Lab Reasoning on distributed data SPARQL
  • 16.
  • 17.
  • 18.
    ELIS  –  Mul*media Lab Sam Coppens, Miel Vander Sande, Ruben Verborgh, Erik Mannens, Rik Van de Walle, Reasoned SPARQL Using SPARQL agents to obtain answers through distribution and reasoning