Human-friendly representations of mapping languages, e.g., YARRRML or ShExML, allow seamlessly integrating heterogeneous data into knowledge graphs. However, they do not describe how a generated knowledge graph should be exported or published, leaving the knowledge graph generation and publication disconnected. To address this, we recently introduced the Logical Target in RML, but an alignment with the human-friendly representations of mapping languages is still pending. In this demo paper, we (i) align RML's Logical Target with YARRRML to provide a more human-friendly syntax for knowledge graph's generation and publication and (ii) demonstrate our approach in YARRRML's editor Matey. YARRRML users are now able to describe how their generated knowledge graphs are exported to one or multiple targets, paving the path towards a more reproducible and human-friendly work ow for generating and publishing knowledge graphs. Demo: https://w3id.org/yarrrml/matey Screencast: https://w3id.org/yarrrml/matey/screencast-target Repository: https://w3id.org/yarrrml/matey/repository
Human-friendly representations of mapping languages, such as YARRRML [
Commons License Attribution 4.0 International (CC BY 4.0).
These human-friendly representations need to follow the evolution of
mapping languages to make all their features available. For instance, we recently
extended RML with the Logical Target to describe how a KG should be exported
to one or multiple targets [
In the case of Logical Target, neither YARRRML nor ShExML, the two most common human-friendly representations of RML, were not extended yet to de ne how and where to export a KG. Consequently, their users still resort to hard-coded scripts to bridge this gap, leaving the KG generation and publication disconnected, even though it is addressed by the underlying mapping language.
In this demo paper, we (i) introduce a YARRRML syntax for RML's Logical
Target1 and (ii) demonstrate our approach in Matey, an editor for YARRRML [
Logical Target alignment with YARRRML
We extend YARRRML's syntax to de ne how and to where a KG, represented in
the RDF [
Targets can be de ned in the same way as sources, both describe how a source or target is accessed (Listing 1.1). YARRRML mappings may refer to targets in subjects, predicates, objects, or graphs maps to describe where RDF triples of the generated KG are exported.
Each target contains at least an access description and type to specify how the target should be accessed (Listing 1.1: lines 5-6, 9). The serialization format and compression are optional (Listing 1.1: lines 7-8). By default, the serialization format is N-Quads and no compression is applied, following the DataIO speci cation2. In Listing 1.1, we de ne: (i) a local le3 to which the triples are dumped with Turtle as serialization format and GZip compression, and (ii) a SPARQL endpoint to which part of the KG is exported to using SPARQL UPDATE. 1 RML's Logical Target alignment is integrated in the YARRRML speci cation: http://w3id.org/yarrrml/spec 2 DataIO speci cation: http://rml.io/specs/dataio 3 VoID Dataset: http://www.w3.org/TR/void Listing 1.1. Targets follow a similar syntax as sources. The SPARQL endpoint target is written in YARRRML's shortcut syntax. 1 sources : 2 people - source : ["/ data / people . json jsonpath ", "$ .[*]"] 3 targets : 4 file - target : 5 access : / data / file . ttl . gz 6 type : void 7 serialization : turtle 8 compression : gzip 9 sparql - target : [" https :// example . org / sparql sparql "]
Each YARRRML mapping may contain a targets map with one or multiple targets to describe where each RDF triple of the generated KG is exported to.
Triples are exported to the targets as speci ed in the mappings (Listing 1.2:
lines 6, 10). If no targets are speci ed for a certain triple, the triple is exported to
the mapping language processor's default target. When multiple targets are
speci ed for the same triple, it is exported to all its targets. This approach allows ne
grained control over where each triple is exported to, following RML's Logical
Target extension [
Listing 1.2. Triples generated by a mapping matching multiple targets are exported to all targets. Triples without a target are exported to the default target. 1 mappings : 2 people : 3 sources : people - source 4 subjects : 5 - value : " http :// example . org / person /$( id )" 6 targets : file - target 7 predicateobjects : 8 - predicates : 9 - value : foaf : name 10 targets : sparql - target 11 objects : "$( name )" 12 - predicates : foaf : age 13 objects : "$( age )" 3
Matey: Human-friendly KG generation & publication We extended YARRRML's editor Matey to demonstrate how a KG is exported to multiple targets such as local les or SPARQL endpoints. We added support for (i) RML's Logical Target extension in YARRRML and (ii) multiple outputs (Figure 1). Each target is shown in Matey in the output tab, the drop-down menu provides the user the possibility to show di erent targets. Matey is available at http://w3id.org/yarrrml/matey and a demo screencast at http://w3id.org/yarrrml/matey/screencast-target which shows how targets are de ned and how this a ects the KG publication. Our extensions to Matey are available under the MIT license: http://w3id.org/yarrrml/matey/repository. Through RML's Logical Target alignment with YARRRML, users of these humanfriendly representations of mapping languages do not need to resort to hardcoded scripts anymore to export their KG to various targets, but they can specify how their generated KG should be exported. Currently, we implemented a few targets such as local les or SPARQL endpoints, but our implementation can be extended to support more targets. Future research is needed to make sure that these human-friendly representations follow the evolution of the languages they represent. In the future, our approach can inspire adjustments to other human-friendly representations of mapping languages, such as ShExML.