1
RDF Mapping Rules Refinements according to Data Consumers’ Feedback RML is a generic language, extending the W3C R2RML language, for defining rules to map data in heterogeneous formats to the RDF data model. RML mappings themselves are graphs. http://semweb.mmlab.be/rml http://semweb.mmlab.be iminds.be Anastasia Dimou, Miel Vander Sande, Tom De Nies, Erik Mannens, Rik Van de Walle {firstname.surname}@ugent.be Ghent University – iMinds – Multimedia Lab <Dep> … <Person> <Name>Bill></Name> <Surname>Gates</Name> <income>10000</income> </Person> … </Dep> <http://ex.com/Gates> a foaf:Person; ex:income "100000". <#PersonMap> rr:subjectMap [ rr:template "http://ex.com/{/Dep/Person/Surname}" ; rr:class foaf:Person ] ; rr:predicateObjectMap [ rr:predicate ex:income ; rr:objectMap [ rml:reference "/Dep/Person/income" ]]. <#PersonMap> rr:subjectMap [ rr:template "http://ex.com/{/Dep/Person/Surname}" ; rr:class foaf:Person ] ; rr:predicateObjectMap [ rr:predicate ex:income ; rr:objectMap [ rml:reference "/Dep/Person/income" rr:datatype xsd:double]] <#PersonMap> rr:subjectMap [ rr:template "http://ex.com/{/Dep/Person/Name}" ; rr:class ex:Person ]; rr:predicateObjectMap [ rr:predicate ex:income ; rr:objectMap [ rml:reference "/Dep/Person/income" ]]. R&Wbase writes data CONSUMERS <http://ex.com/Bill> a ex:Person ; ex:income "100000". reads data OWNERS maps generates refines Primary resource maps new version <Dep> … <Person> <Name>Bill></Name> <Surname>Gates</Name> <income>20000</income> </Person> … </Dep> Primary resource generates <http://ex.com/Gates> a foaf:Person; ex:income "200000". <http://ex.com/Gates> ex:income "200000"^^xsd:double. refines PROV-O PROV-O PROV-O PROV-O reads writes Applications Users A triple is generated by the data owner. The triple is modified by a data consumer. The triple’s provenance information (PROV-O) indicates the RML mapping that generated it. The RML mapping is refined according to the data consumer’s modifications. The next time an updated version of the primary source is mapped, the updated RML mapping applies also the data consumer’s refinements. Published Linked Data RML mapping rules R&Wbase is an interpretation layer built on top of SPARQL endpoints that stores different deltas in a single graph, allowing simple resolving of different versions. R&Wbase can track all changes made to the generated triples. http://rawbase.github.io

RDF Mapping Rules Refinements According to Data Consumers' Feedback

  • Upload
    andimou

  • View
    78

  • Download
    6

Embed Size (px)

DESCRIPTION

The missing feedback loop is considered the reason for broken Data Cycles on current Linked Open Data ecosystems. Read-Write platforms are proposed, but they are restricted to capture modifications after the data is released as Linked Data. Triggering though a new iteration results in loosing the data consumers’ modifications, as a new version of the source data is mapped, overwriting the currently published. We propose a prime solution that interprets the data consumers’ feedback to update the mapping rules. This way, data publishers initiate a new iteration of the Data Cycle considering the data consumers’ feedback when they map a new version of the published data.

Citation preview

Page 1: RDF Mapping Rules Refinements According to Data Consumers' Feedback

RDF Mapping Rules Refinements according to Data Consumers’ Feedback

RML is a generic language, extending the

W3C R2RML language, for defining rules to

map data in heterogeneous formats to the

RDF data model. RML mappings themselves

are graphs. http://semweb.mmlab.be/rml

http://semweb.mmlab.be • iminds.be

Anastasia Dimou, Miel Vander Sande, Tom De Nies, Erik Mannens, Rik Van de Walle

{firstname.surname}@ugent.be Ghent University – iMinds – Multimedia Lab

<Dep> …

<Person>

<Name>Bill></Name>

<Surname>Gates</Name>

<income>10000</income>

</Person>

… </Dep>

<http://ex.com/Gates>

a foaf:Person;

ex:income "100000".

<#PersonMap>

rr:subjectMap [

rr:template "http://ex.com/{/Dep/Person/Surname}" ;

rr:class foaf:Person ] ;

rr:predicateObjectMap [ rr:predicate ex:income ;

rr:objectMap [ rml:reference "/Dep/Person/income" ]].

<#PersonMap>

rr:subjectMap [

rr:template "http://ex.com/{/Dep/Person/Surname}" ;

rr:class foaf:Person ] ;

rr:predicateObjectMap [

rr:predicate ex:income ;

rr:objectMap [ rml:reference "/Dep/Person/income"

rr:datatype xsd:double]]

<#PersonMap>

rr:subjectMap [

rr:template "http://ex.com/{/Dep/Person/Name}" ;

rr:class ex:Person ];

rr:predicateObjectMap [ rr:predicate ex:income ;

rr:objectMap [ rml:reference "/Dep/Person/income" ]].

R&Wbase

writes

da

ta C

ON

SU

ME

RS

<http://ex.com/Bill>

a ex:Person ;

ex:income "100000".reads

da

ta O

WN

ER

S

mapsgenerates

refines

Primary resource

maps

new

version

<Dep> …

<Person>

<Name>Bill></Name>

<Surname>Gates</Name>

<income>20000</income>

</Person>

… </Dep>

Primary resource

generates<http://ex.com/Gates>

a foaf:Person;

ex:income "200000".

<http://ex.com/Gates>

ex:income

"200000"^^xsd:double.

refines

PROV-O

PROV-O

PROV-O

PROV-O

reads

writesAp

pli

cati

on

sU

sers

A triple is generated by the data owner.

The triple is modified by a data consumer.

The triple’s provenance information (PROV-O)

indicates the RML mapping that generated it.

The RML mapping is refined according to

the data consumer’s modifications.

The next time an updated version of the primary

source is mapped, the updated RML mapping

applies also the data consumer’s refinements.

Published

Linked Data

RML

mapping rules

R&Wbase is an interpretation layer built on

top of SPARQL endpoints that stores

different deltas in a single graph, allowing

simple resolving of different versions.

R&Wbase can track all changes made to the

generated triples. http://rawbase.github.io