[Debwebid-discuss] rdf coordination

[chrysn]

Hi!

On Wed, Aug 12, 2015 at 11:25:01PM +0100, Iain R. Learmonth wrote:
> If the URIs are not dereferencable, they're useless as far as I'm concerned.

That's what I mean by being nice participants of the semantic web, to be
useful.

> > http://packages.debian.org/src:arandr#project
> 
> Not dereferenceable and depends on packages.debian.org not polluting that
> namespace which cannot be guaranteed at present.

Not currently, but a negotiating redirect to whatever serves the actual
metadata should be easy to implement.

I've given the meta.debian.net some further thought; I do agree that it
is easier to manage the namespace if it were under control of RDF
afficionados.

I think I can express more clearly now why I still prefer the
related-service approach: Debian is a diverse project, with
interconnected systems that are managed quite separately. Using the
namespace of whichever subsystem the resource we describe belongs to
means working with more services, but reflects that spirit better.

On the practical side, this allows starting a new service more easily.
Were debtags launched after our considerations, its maintainer could set
up the debtags.debian.net on its own and start serving RDF data that
interlinks well without going through UDD and meta.d.o.

For demonstrating semantic technologies, I think this would work well
too: A SPARQL endpoint or a full-debian-data triple dump would just need
to have additional sources (like the example above) added once they've
stabilized, without any single entity needing to have the whole model
in view.

Of course, the UDD is practical for this, and might easily serve as data
source for an implementation for what semantic-debian currently serves.
For example, http://packages.debian.org/src:arandr might 303 to
http://rdf.debian.net/project/arandr . (The rdf. name might even stay,
then -- we're talking about "The doap:project that comprises maintaining
the arandr soure package in Debian", and for the given request, data is
served by our RDF metadata provider).

Is this a line of thought you can relate to?


> > Summarising, the question I'd like feedback on in this thread is: What
> > is the most suitable URI, for example, for "the arandr source package"?
> 
> Are you talking about the abstract concept of the source package, the files
> that comprise it or the project that maintains the package? Are you talking
> about a specific version of the package or a collection of all versions?

Both current data exports have one version-independent resource for each
source package name (like the above) and their individual releases; I
think that's uncontroversial.

> I am considering dropping ADMSSW and going with pure DOAP which may simplify
> things, but this is still something that needs more thought (and I have
> given it a lot so far).

I don't have an opinion on that so far; ADMSSW seems to properly
subclass/-property DOAP, and it might or might not add important
distinctions.


Best regards
chrysn

-- 
To use raw power is to make yourself infinitely vulnerable to greater powers.
  -- Bene Gesserit axiom
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 819 bytes
Desc: Digital signature
URL: <http://lists.alioth.debian.org/pipermail/debwebid-discuss/attachments/20150813/7cb45d2f/attachment.sig>