Datalink Feedback VI: Semantics

[Markus Demleitner]

Hi all,

On Wed, Apr 23, 2014 at 04:29:23PM +0100, Norman Gray wrote:
> The following email is quite 'bitty', consisting of rejoinders to
> Markus's points rather than assembling a continuous argument.

In short -- I give in.  While I'm not convinced going proper RDF is
all that beneficial and carries no cost, I really want to avoid
leaving semantics unspecified in datalink 1.0.  In particular, being
able to identify the dataset itself is very important to our client
implementation (e.g., on the CADC Obscore service that only hands out
datalink now), so that needs to be solidly defined or something
fairly central will be broken.

Just two things:

> It doesn't mean that everyone has to understand all of the terms.

... but it would be very nice if it were easy for a client to
understand all the terms that are likely in use.  Hence, building in
a strong incentive to just use the IVOA-defined terms (and maybe we
can improve the glacier-scale speeds there rather than work around
them) would, I think, be highly useful.

> To the extent that I'm proposing a Plan, I suppose it'd be this:
> 
>   1. the items in the Datalink 'semantics' column are URLs which
>   are taken to be RDF Properties
> 
>   2. the language in Sect. 3.2.6 of the Datalink spec indicates in
>   a sentence that each row of this list of links (which has a
>   'semantics' entry) is interpretable  as an RDF triple
> 
>   3. any URL is permitted in the 'semantics' column, and clients
>   are not required to understand any of them, but...
> 
>   4. there is a parallel Datalink Vocabulary document, in
>   preparation, which defines some 'blessed' ones. which it would be
>   silly for a client not to understand.

Please, no, let's not say "in preparation".  Either the vocabulary
comes with datalink, or everyone will be horribly confused.  And
also, let's avoid having long URLs with 90% of data data constant in
semantics by default.  Unwieldyness matters even when users shouldn't
normally be seeing these things.  Can't we say something like:

  The values of the semantics column are URLs.  If semantics just
  contains a fragment ("#term"), the fragment refers to
  http://ivoa.net/rdf/datalink [or whatever].  Clients should
  understand at least all the values mentioned in that vocabulary,
  and they should interpret the rdfs:subPropertyOf relations
  mentioned in there.

Added benefit: These things will look like hashtags.

Uh... <bambi eyes>... Norman -- since that's probably easiest for you
among everyone that's reading this: Do you think you could draft
such a vocabulary, maybe based on Francois' suggestions
http://www.ivoa.net/pipermail/dal/2014-April/006781.html (plus
"self", which I suppose would either be top-level or under something
like "alternative representation")?  As I said, I'm convinced if we
want this to fly and be understood by the deployers, the vocabulary
will have to come with datalink.

> Re 1: I think that 'relation' or 'predicate' would be a better name
> than 'semantics'.  Also, the idea of an RDF 'Property' (aka
> 'Predicate') is simply that it's the link between a 'subject' and
> an 'object':

If we go for a formal RDF interpretation, I'm all for calling the
thing property (except if that happened to be a reserved word in some
RDBMS...).

Cheers,

       Markus