search

opengeospatial / ogc-geosparql

Public Repository for the OGC GeoSPARQL Standards Working Group
77 stars 20 forks source link

consider implications of RDF* and SPARQL* #34

Open rob-metalinkage opened 4 years ago

rob-metalinkage commented 4 years ago

a big challenge for geometry is the metadata needed to qualify it.

There seem to be a few approaches: 1) proliferation of properties nuanced to classifications of qualification 2) embedded qualifications in a syntax inside data type serialisations (... some form of WKT) 3) reification of RDF statements 4) naming conventions for groups properties containing different aspects of qualification 5) changing data models to always use qualified associations 6) native support for properties of properties - RDF

RDF* is probably the simplest option here - but it could be defined with an equivalent RDF reification pattern and/or qualified association

https://blog.liu.se/olafhartig/2019/01/10/position-statement-rdf-star-and-sparql-star/

mathib commented 4 years ago

This is a useful source (but already from 2015) comparing different approaches (standard reification, n-ary relations, named graphs, singleton properties) for statement metadata: http://aidanhogan.com/docs/reification-wikidata-rdf-sparql.pdf. Regarding the support at the time of the research, n-ary relations and standard reification were clearly better options for statement metadata. Of the last two methods, I think n-ary relations are preferred since they are more intuitive to model.

RDF/SPARQL are not on a standards track as far as I know (there might come a W3C member submission note though). The support for both of them is still rather low (Blazegraph, GraphDB, Stardog and Anzograph). Therefore, I would turn the last sentence a bit around: N-ary relations (= qualified association?) are currently the most straightforward and well-supported approach for adding metadata. In addition, we can define RDF* and RDF reificiation representations.

jabhay commented 4 years ago

Created MoSCoW poll

FransKnibbe commented 4 years ago

I voted C here, because on the one hand I'm all for considering implications of all new ideas for improving the semantic web. On the other hand, I don't see the problem of using properties of geometries to make things clear. I'm am not sure the problem of making statements about RDF statements applies to publishing geometric data.

akuckartz commented 3 years ago

The main website about "RDF-star" (less than a week ago it was decided that this is a better name than "RDF*") is: https://w3c.github.io/rdf-star/