W3C home > Mailing lists > Public > public-dxwg-wg@w3.org > July 2020

Re: [dxwg] Question: how to catalog relational database data in DCAT? (#1240)

From: Rob Atkinson via GitHub <sysbot+gh@w3.org>
Date: Sat, 11 Jul 2020 06:53:15 +0000
To: public-dxwg-wg@w3.org
Message-ID: <issue_comment.created-657003939-1594450394-sysbot+gh@w3.org>
Hi @zeginis -
(I am also working for the OGC in the dissemination workpackage for CYBELE, so availalable to help with any promotion (and testing ;-) ) of candidate best practices.

 in order to describe data well enough to be able to query it via a service you need at least five different things:
1) location of a service endpoint 
2) description of the service API - whether it be SPARQL, OGC WFS, OGCAPI, NGSI-LD or any other thing
3) description of the data schema (what elements will be returned for a query - aka FeatureType in an OGC services
4) description of queryable aspects of data - either data dimensions or parameters of query functions cover most if not all cases here
5) knowledge of the allowable range of queryable aspects (spatio-temporal extent, codelists etc)

DCAT provides for 1, and using dcterms:conformsTo to identify serviceType can at least identify any self-descriptive capabilities of the service - such as OAS or OGC GetCapabilities.  Any more detail and you need to define your own profile of DCAT with additional metadata properties needed to describe the service.

Some services can describe the data schema - e.g. WFS describeFeatureType - but AFAIK nothing in widespread use does a reasonable job for #4 and 5.

The suggested best practice from the statistics community id the use of RDF-Datacube vocabulary to handle 4 and 5.  There is a W3C/OGC Note describing a possible spatio-temporal profile of RDF-Datacube called QB4ST [https://www.w3.org/TR/qb4st/] which directly addresses this gap, but to date little effort has been put into semantic description of query interfaces or even data services.  This requires testing in live context and I'd be very happy to assist you with the general challenge of creating expressive enough metadata using available standards.

The one item I know needs to be addressed to achieve a complete solution is the bridge between RDF-Datacube which allows for description of rdf:Property elements, and description of relational database (or JSON, XML or any other meta-model).   This  needs either:
*  specialised properties for different forms of property reference such as something like qbex:xpath (for the case of XPath elements identifiers)
* a qualified association to an property description object, where the property description object can declare what type of description it is, and this can be extended with any form of element identifier we might need in future.

Lets talk about how to achieve this and explore any support such as publishig formalised profiles of DCAT that can assist.

GitHub Notification of comment by rob-metalinkage
Please view or discuss this issue at https://github.com/w3c/dxwg/issues/1240#issuecomment-657003939 using your GitHub account
Received on Saturday, 11 July 2020 06:53:20 UTC

This archive was generated by hypermail 2.4.0 : Saturday, 11 July 2020 06:53:27 UTC