RE: BP restructuring: Data Quality and Versioning sections

Hi,

I don’t think this is a bad time for this. Restructuring and alignment with DWBP may also bring things like this to light!

I think dataset-level metadata is a good thing to have, among other things, for discoverability. While being able to discover entities from a dataset directly on the web (i.e. crawlable data) is useful, actually finding individual entities may baffle people; people may be able to judge much more easily from dataset metadata if the data is useful for their purposes. Hence I think making dataset metadata crawlable is also important. Not in an obscure format, but in format(s) understandable by web crawlers and web developers.

What kind of metadata are you talking about? Are there any hooks in the use cases or requirements we’ve defined in the UCR?

Things like CRS, UoM and precision can be relevant at property level but are very often the same for an entire dataset. (although I have a use case where accuracy isn’t)

Your proposal for a BP “provide explicit machine-readable  metadata for spatial properties” where dataset level metadata is one of the options could work, though. What do other WG members think?

Linda

Van: Rob Atkinson [mailto:rob@metalinkage.com.au]
Verzonden: dinsdag 2 augustus 2016 01:02
Aan: Linda van den Brink; SDW WG (public-sdw-wg@w3.org)
Onderwerp: Re: BP restructuring: Data Quality and Versioning sections


Hi,

I'd like to look at the BP to address its basic implementability for the cases where machine-readable metadata is needed for attributes - as is the common case for spatial, but also  more generally applicable cases (CRS, UoM, Precision).  I'm trying to work out when the dust has settled enough from the restructure to make change proposals. Reading it through I am struck by a few concerns that it may be harder if left too long...

In particular I dont feel sufficient acknowledgement is made of the basic issue - metadata is necessary - but on the Web metadata about datasets is less important than being able to interpret data you get back from it, or the access methods you need to get it. Fine to have a section about traditional dataset metadata and point to geo-dcat ( a nice strong recommendation that is useful).  Less good IMHO is to make references back to this, implicitly or explicity as a preferred mechanism - for most of the Web its a corner case - in reality no web developer is likely to find data and try to automate access and interpretation of a metadata record in an obscure format (for them XML!)

IMHO we need to have a section about attribute level spatial metadata that explicitly states it may be embedded in many places - provide some vanilla statements about choice of appropriate vocabularies and code lists with URI. This section then refer to dataset metadata _as one option_ and point to that section, but also provide options for embedding this metadata into data itself.  We can point back to this every time its relevant, rather than a dataset-specific metadata section (BP1) or an even vaguer reference to DCAT which kind of implies it, as its the only place it make sense.

I dont think this is a particularly hard thing to do, but for me its a BP0  - provide explicit machine-readable  metadata for spatial properties, where machine-readable includes using a URI identifier for the spatial property (e.g. geo:latitude), or additional properties using vocab X (can we recommend one - or reference a note describing an option?) with URI or embedded data structures.

Every place where a BP demands such annotations, we can point to this, with its embed-or-describe-with-metadata-record pattern, and achieve greater flexibility, readability and consistency of practice.

Rob



On Mon, 1 Aug 2016 at 22:58 Linda van den Brink <l.vandenbrink@geonovum.nl<mailto:l.vandenbrink@geonovum.nl>> wrote:
Hi all,

At this point I welcome more detailed comments on the section in the BP on the Spatial Data Quality and Spatial Data Versioning sections.
- http://w3c.github.io/sdw/bp/#bp-dataquality

- http://w3c.github.io/sdw/bp/#bp-dataversioning


Spatial Data Quality already has some pretty good content, including examples, thanks to Andrea who helped me with this.

Spatial Data Versioning is more drafty. Here I could do with any comments, corrections or additions, including feedback on the two issues mentioned in the text. Examples are also still missing.

Linda

______________________________________
Geonovum
Linda van den Brink
Adviseur Geo-standaarden

a: Barchman Wuytierslaan 10, 3818 LH Amersfoort
p: Postbus 508, 3800 AM Amersfoort
t:  + 31 (0)33 46041 00
m: + 31 (0)6 1355 57 92
e:  l.vandenbrink@geonovum.nl<mailto:r.beltman@geonovum.nl>
i:  www.geonovum.nl<http://www.geonovum.nl/>
tw: @brinkwoman