Re: Geographic metadata in Spatial Data on the Web Best Practices

Another perspective - spatial data on the web - in any form - requires rich
and flexible metadata to make it discoverable, invokable and interpretable
- and a "semantic web" is really the only candidate for interoperable,
canonical forms of this metadata.I say this because there are already some
useful components available for re-use - and the meta-model supports the
key multi-inheritance patterns we will care about when describing resource
behaviours from multiple user perspectives.

Linked Data provides a minimalist approach to this - linking using
dereferencerable URIs it can be implemented agnostically of the encoding -
HTML, RDF/XML, RDF-TTL, JSON-LD etc. IMHO it allows an incremental
development of interoperability without needing to finalise a single shared
information model.

If we choose the Web - then we are also really choosing URI based
vocabularies for key concepts - and the question is what are the minimal
set of concepts we need to define to support some specific level of
interoperability between these notional metadata graphs. IMHO the business
of making the data itself interoperable is handled elsewhere by defining
domain models and encodings - and these may or may not involve RDF - so the
critical part of the SDW scope really is all about discovery and linking.

That said, there is also a part about defining vocubularies for
spatio/temporal concepts - and these should naturally be common across the
data and metadata - so a semweb-oriented approach to these is a short-term
enabler - but doesnt necessarily mean that SDW needs to proscribe data as
RDF as the only way forward.

Rob Atkinson


On Mon, 7 Mar 2016 at 23:14 Erik Wilde <erik.wilde@dret.net> wrote:

> hello neil.
>
> On 2016-03-07 10:02, Neil McNaughton wrote:
> > /Another comment – there is no mention of the semantic web. Has this to
> > all intents and purposes been replaced by “Linked Data?” Is this just a
> > buzzword swap or has something more substantial happened? I ask because
> > the ‘payload’ of the /Spatial Data on the Web Best Practices document
> > appears to be relations mapped in “OWL, SKOS, RDFS” which to the casual
> > observer sounds like the semantic web.//
>
> yes, "linked data" is a rebranding of "semantic web", plus a few
> additional constraints (most importantly: "use dereferencable HTTP URIs
> for everything").
>
> fyi, there have been discussions on whether SDW should be RDF-centric or
> not. some (including myself) have argued that "the web" is much wider
> than the "semantic web", and that the draft in its current form should
> either be titled "spatial data on the semantic web", or should be
> changed to be agnostic of a specific metamodel and simply recommend
> patterns and best practices derived from web architecture.
>
> to this end, http://dret.github.io/webdata/ is something that could
> serve as a foundation or starting point: it talks about the principles
> of web architecture without mandating one specific metamodel. it's
> basically "linked data minus requiring RDF".
>
> this issue of "the current BP draft is for semweb users only" has been
> raised before. it remains to be seen which path the WG and the spec are
> going to take.
>
> cheers,
>
> dret.
>
> --
> erik wilde | mailto:erik.wilde@dret.net |
>             | http://dret.net/netdret    |
>             | http://twitter.com/dret    |
>
>